Giving Your AI Agents Reliable Skills with the Agent Skills SDK
AI agents are becoming increasingly capable, but they often do not have the context they need to do real work reliably. Your agent can reason well, but it does not actually know how to do the specific things your team needs it to do. For example, it cannot follow your company's incident response playbook, it does not know your escalation policy, and it has no idea how to page the on-call engineer at 3 AM. There are many ways to close this gap, from RAG to custom tool implementations. Agent Skills is one approach that stands out because it is designed around portability and progressive disclosure, keeping context window usage minimal while giving agents access to deep expertise on demand. What is Agent Skills? Agent Skills is an open format for giving agents new capabilities and expertise. The format was originally developed by Anthropic and released as an open standard. It is now supported by a growing list of agent products including Claude Code, VS Code, GitHub, OpenAI Codex, Cursor, Gemini CLI, and many others. As defined in the spec, a skill is a folder on disk containing a SKILL.md file with metadata and instructions, plus optional scripts, references, and assets: incident-response/ SKILL.md # Required: instructions + metadata references/ # Optional: additional documentation severity-levels.md escalation-policy.md scripts/ # Optional: executable code page-oncall.sh assets/ # Optional: templates, diagrams, data files The SKILL.md file has YAML frontmatter with a name and description (so agents know when the skill is relevant), followed by markdown instructions that tell the agent how to perform the task. The format is intentionally simple: self-documenting, extensible, and portable. What makes this design practical is progressive disclosure. The spec is built around the idea that agents should not load everything at once. It works in three stages: Discovery: At startup, agents load only the name and description of each available skill, just enough to know when it might be relevant. Activation: When a task matches a skill's description, the agent reads the full SKILL.md instructions into context. Execution: The agent follows the instructions, optionally loading referenced files or executing bundled scripts as needed. This keeps agents fast while giving them access to deep context on demand. The format is well-designed and widely adopted, but if you want to use skills from your own agents, there is a gap between the spec and a working implementation. The Agent Skills SDK Conceptually, a skill is more than a folder. It is a unit of expertise: a name, a description, a body of instructions, and a set of supporting resources. The file layout is one way to represent that, but there is nothing about the concept that requires a filesystem. The Agent Skills SDK is an open-source Python library built around that idea, treating skills as abstract units of expertise that can be stored anywhere and consumed by any agent framework. It does this by addressing two challenges that come up when you try to use the format from your own agents. The first is where skills live. The spec defines skills as folders on disk, and the tools that support the format today all assume skills are local files. Files are inherently portable, and that is one of the format's strengths. But in the real world, not every team can or wants to serve skills from the filesystem. Maybe your team keeps them in an S3 bucket. Maybe they are in Azure Blob Storage behind your CDN. Maybe they live in a database alongside the rest of your application data. At the moment, if your skills are not on the local filesystem, you are on your own. The SDK changes where skills are served from, not how they are authored. The content and format stay the same regardless of the storage backend, so skills remain portable across providers. The second is how agents consume them. The spec defines the progressive disclosure pattern but actually implementing it in your agent requires real work. You need to figure out how to validate skills against the spec, generate a catalog for the system prompt, expose the right tools for on-demand content retrieval, and handle the back-and-forth of the agent requesting metadata, then the body, then individual references or scripts. That is a lot of plumbing regardless of where the skills are stored, and the work multiplies if you want to support more than one agent framework. The SDK solves both by separating where skills come from (providers) from how agents use them (integrations), so you can mix and match freely. Load skills from the filesystem today, move them to an HTTP server tomorrow, swap in a custom database provider next month, and your agent code does not change at all. How the SDK works The SDK is a set of Python packages organized around two ideas: storage-agnostic providers and progressive disclosure. The provider abstraction means your skills can live anywhere. The SDK ships with providers for the local filesystem and static HTTP servers, but the SkillProvider interface is simple enough that you can write your own in a few methods. A Cosmos DB provider, a Git provider, a SharePoint provider, whatever makes sense for your team. The rest of the SDK does not care where the data comes from. On top of that, the SDK implements the progressive disclosure pattern from the spec as a set of tools that any LLM agent can use. At startup, the SDK generates a skills catalog containing each skill's name and description. Your agent injects this catalog into its system prompt so it knows what is available. Then, during a conversation, the agent calls tools to retrieve content on demand, following the same discovery-activation-execution flow the spec describes. Here is the flow in practice: You register skills from any source (local files, an HTTP server, your own database). The SDK generates a catalog and tool usage instructions, which you inject into the system prompt. The agent calls tools to retrieve content on demand. This matters because context windows are finite. An incident response skill might have a main body, three reference documents, two scripts, and a flowchart. The agent should not load all of that upfront. It should read the body first, then pull the escalation policy only when the conversation actually gets to escalation. A quick example Here is what it looks like in practice. Start by loading a skill from the filesystem: from pathlib import Path from agentskills_core import SkillRegistry from agentskills_fs import LocalFileSystemSkillProvider provider = LocalFileSystemSkillProvider(Path("my-skills")) registry = SkillRegistry() await registry.register("incident-response", provider) Now wire it into a LangChain agent: from langchain.agents import create_agent from agentskills_langchain import get_tools, get_tools_usage_instructions tools = get_tools(registry) skills_catalog = await registry.get_skills_catalog(format="xml") tool_usage_instructions = get_tools_usage_instructions() system_prompt = ( "You are an SRE assistant. Use the available skill tools to look up " "incident response procedures, severity definitions, and escalation " "policies. Always cite which reference document you used.\n\n" f"{skills_catalog}\n\n" f"{tool_usage_instructions}" ) agent = create_agent( llm, tools, system_prompt=system_prompt, ) That is it. The agent now knows what skills are available and has tools to fetch their content. When a user asks "How do I handle a SEV1 incident?", the agent will call get_skill_body to read the instructions, then get_skill_reference to pull the severity levels document, all without you writing any of that retrieval logic. The same pattern works with Microsoft Agent Framework: from agentskills_agentframework import get_tools, get_tools_usage_instructions tools = get_tools(registry) skills_catalog = await registry.get_skills_catalog(format="xml") tool_usage_instructions = get_tools_usage_instructions() system_prompt = ( "You are an SRE assistant. Use the available skill tools to look up " "incident response procedures, severity definitions, and escalation " "policies. Always cite which reference document you used.\n\n" f"{skills_catalog}\n\n" f"{tool_usage_instructions}" ) agent = Agent( client=client, instructions=system_prompt, tools=tools, ) What is in the SDK The SDK is split into small, composable packages so you only install what you need: agentskills-core handles registration, validation, the skills catalog, and the progressive disclosure API. It also defines the SkillProvider interface that all providers implement. agentskills-fs and agentskills-http are the two built-in providers. The filesystem provider loads skills from local directories. The HTTP provider loads them from any static file host: S3, Azure Blob Storage, GitHub Pages, a CDN, or anything that serves files over HTTP. agentskills-langchain and agentskills-agentframework generate framework-native tools and tool usage instructions from a skill registry. agentskills-mcp-server spins up an MCP server that exposes skill tool access and usage as tools and resources, so any MCP-compatible client can use them. Because providers and integrations are separate packages, you can combine them however you want. Use the filesystem provider during development, switch to the HTTP provider in production, or write a custom provider that reads skills from your own database. The integration layer does not need to know or care. Where to go from here The full source, working examples, and detailed API docs are on GitHub: github.com/pratikxpanda/agentskills-sdk The repo includes end-to-end examples for both LangChain and Microsoft Agent Framework, covering filesystem providers, HTTP providers, and MCP. There is also a sample incident-response skill you can use to try things out. A proposal to contribute this SDK to the official agentskills repository has been submitted. If you find it useful, feel free to show your support on the GitHub issue. To learn more about the Agent Skills format itself: What are skills? covers the format and why it matters. Specification is the complete format reference for SKILL.md files. Integrate skills explains how to add skills support to your agent. Example skills on GitHub are a good starting point for writing your own. The SDK is MIT licensed and contributions are welcome. If you have questions or ideas, post a question here or open an issue on the repo.
Exploring Azure Face API: Facial Landmark Detection and Real-Time Analysis with C#
In today’s world, applications that understand and respond to human facial cues are no longer science fiction—they’re becoming a reality in domains like security, driver monitoring, gaming, and AR/VR. With Azure Face API, developers can leverage powerful cloud-based facial recognition and analysis tools without building complex machine learning models from scratch. In this blog, we’ll explore how to use C# to detect faces, identify key facial landmarks, estimate head pose, track eye and mouth movements, and process real-time video streams. Using OpenCV for visualization, we’ll show how to overlay landmarks, draw bounding boxes, and calculate metrics like Eye Aspect Ratio (EAR) and Mouth Aspect Ratio (MAR)—all in real time. You'll learn to: Set up Azure Face API Detect 27 facial landmarks Estimate head pose (yaw, pitch, roll) Calculate eye aspect ratio (EAR) and mouth openness Draw bounding boxes around features using OpenCV Process real-time video Prerequisites .NET 8 SDK installed Azure subscription with Face API resource Visual Studio 2022 or later Webcam for testing (optional) Basic understanding of C# and computer vision concepts Part 1: Azure Face API Setup 1.1 Install Required NuGet Packages dotnet add package Azure.AI.Vision.Face dotnet add package OpenCvSharp4 dotnet add package OpenCvSharp4.runtime.win 1.2 Create Azure Face API Resource Navigate to Azure Portal Search for "Face" and create a new Face API resource Choose your pricing tier (Free tier: 20 calls/min, 30K calls/month) Copy the Endpoint URL and API Key 1.3 Configure in .NET Application appsettings.json: { "Azure": { "FaceApi": { "Endpoint": "https://your-resource.cognitiveservices.azure.com/", "ApiKey": "your-api-key-here" } } } Initialize Face Client: using Azure; using Azure.AI.Vision.Face; using Microsoft.Extensions.Configuration; public class FaceAnalysisService { private readonly FaceClient _faceClient; private readonly ILogger<FaceAnalysisService> _logger; public FaceAnalysisService(ILogger<FaceAnalysisService> logger, IConfiguration configuration) { _logger = logger; string endpoint = configuration["Azure:FaceApi:Endpoint"]; string apiKey = configuration["Azure:FaceApi:ApiKey"]; _faceClient = new FaceClient(new Uri(endpoint), new AzureKeyCredential(apiKey)); _logger.LogInformation("FaceClient initialized with endpoint: {Endpoint}", endpoint); } } Part 2: Understanding Face Detection Models 2.1 Basic Face Detection public async Task<List<FaceDetectionResult>> DetectFacesAsync(byte[] imageBytes) { using var stream = new MemoryStream(imageBytes); var response = await _faceClient.DetectAsync( BinaryData.FromStream(stream), FaceDetectionModel.Detection03, FaceRecognitionModel.Recognition04, returnFaceId: false, returnFaceAttributes: new FaceAttributeType[] { FaceAttributeType.HeadPose }, returnFaceLandmarks: true, returnRecognitionModel: false ); _logger.LogInformation("Detected {Count} faces", response.Value.Count); return response.Value.ToList(); } Part 3: Facial Landmarks - The 27 Key Points 3.1 Understanding Facial Landmarks 3.2 Accessing Landmarks in Code public void PrintLandmarks(FaceDetectionResult face) { var landmarks = face.FaceLandmarks; if (landmarks == null) { _logger.LogWarning("No landmarks detected"); return; } // Eye landmarks Console.WriteLine($"Left Eye Outer: ({landmarks.EyeLeftOuter.X}, {landmarks.EyeLeftOuter.Y})"); Console.WriteLine($"Left Eye Inner: ({landmarks.EyeLeftInner.X}, {landmarks.EyeLeftInner.Y})"); Console.WriteLine($"Left Eye Top: ({landmarks.EyeLeftTop.X}, {landmarks.EyeLeftTop.Y})"); Console.WriteLine($"Left Eye Bottom: ({landmarks.EyeLeftBottom.X}, {landmarks.EyeLeftBottom.Y})"); // Mouth landmarks Console.WriteLine($"Upper Lip Top: ({landmarks.UpperLipTop.X}, {landmarks.UpperLipTop.Y})"); Console.WriteLine($"Under Lip Bottom: ({landmarks.UnderLipBottom.X}, {landmarks.UnderLipBottom.Y})"); // Nose landmarks Console.WriteLine($"Nose Tip: ({landmarks.NoseTip.X}, {landmarks.NoseTip.Y})"); } 3.3 Visualizing All Landmarks public void DrawAllLandmarks(FaceLandmarks landmarks, Mat frame) { void DrawPoint(FaceLandmarkCoordinate point, Scalar color) { if (point != null) { Cv2.Circle(frame, new Point((int)point.X, (int)point.Y), radius: 3, color: color, thickness: -1); } } // Eyes (Green) DrawPoint(landmarks.EyeLeftOuter, new Scalar(0, 255, 0)); DrawPoint(landmarks.EyeLeftInner, new Scalar(0, 255, 0)); DrawPoint(landmarks.EyeLeftTop, new Scalar(0, 255, 0)); DrawPoint(landmarks.EyeLeftBottom, new Scalar(0, 255, 0)); DrawPoint(landmarks.EyeRightOuter, new Scalar(0, 255, 0)); DrawPoint(landmarks.EyeRightInner, new Scalar(0, 255, 0)); DrawPoint(landmarks.EyeRightTop, new Scalar(0, 255, 0)); DrawPoint(landmarks.EyeRightBottom, new Scalar(0, 255, 0)); // Eyebrows (Cyan) DrawPoint(landmarks.EyebrowLeftOuter, new Scalar(255, 255, 0)); DrawPoint(landmarks.EyebrowLeftInner, new Scalar(255, 255, 0)); DrawPoint(landmarks.EyebrowRightOuter, new Scalar(255, 255, 0)); DrawPoint(landmarks.EyebrowRightInner, new Scalar(255, 255, 0)); // Nose (Yellow) DrawPoint(landmarks.NoseTip, new Scalar(0, 255, 255)); DrawPoint(landmarks.NoseRootLeft, new Scalar(0, 255, 255)); DrawPoint(landmarks.NoseRootRight, new Scalar(0, 255, 255)); DrawPoint(landmarks.NoseLeftAlarOutTip, new Scalar(0, 255, 255)); DrawPoint(landmarks.NoseRightAlarOutTip, new Scalar(0, 255, 255)); // Mouth (Blue) DrawPoint(landmarks.UpperLipTop, new Scalar(255, 0, 0)); DrawPoint(landmarks.UpperLipBottom, new Scalar(255, 0, 0)); DrawPoint(landmarks.UnderLipTop, new Scalar(255, 0, 0)); DrawPoint(landmarks.UnderLipBottom, new Scalar(255, 0, 0)); DrawPoint(landmarks.MouthLeft, new Scalar(255, 0, 0)); DrawPoint(landmarks.MouthRight, new Scalar(255, 0, 0)); // Pupils (Red) DrawPoint(landmarks.PupilLeft, new Scalar(0, 0, 255)); DrawPoint(landmarks.PupilRight, new Scalar(0, 0, 255)); } Part 4: Drawing Bounding Boxes Around Features 4.1 Eye Bounding Boxes /// <summary> /// Draws rectangles around eyes using OpenCV. /// </summary> public void DrawEyeBoxes(FaceLandmarks landmarks, Mat frame) { int boxWidth = 60; int boxHeight = 35; // Calculate Rectangles var leftEyeRect = new Rect((int)landmarks.EyeLeftOuter.X - boxWidth / 2, (int)landmarks.EyeLeftOuter.Y - boxHeight / 2, boxWidth, boxHeight); var rightEyeRect = new Rect((int)landmarks.EyeRightOuter.X - boxWidth / 2, (int)landmarks.EyeRightOuter.Y - boxHeight / 2, boxWidth, boxHeight); // Draw Rectangles (Green in BGR) Cv2.Rectangle(frame, leftEyeRect, new Scalar(0, 255, 0), 2); Cv2.Rectangle(frame, rightEyeRect, new Scalar(0, 255, 0), 2); // Add Labels Cv2.PutText(frame, "Left Eye", new Point(leftEyeRect.X, leftEyeRect.Y - 5), HersheyFonts.HersheySimplex, 0.4, new Scalar(0, 255, 0), 1); Cv2.PutText(frame, "Right Eye", new Point(rightEyeRect.X, rightEyeRect.Y - 5), HersheyFonts.HersheySimplex, 0.4, new Scalar(0, 255, 0), 1); } 4.2 Mouth Bounding Box /// <summary> /// Draws rectangle around mouth region. /// </summary> public void DrawMouthBox(FaceLandmarks landmarks, Mat frame) { int boxWidth = 80; int boxHeight = 50; // Calculate center based on the vertical lip landmarks int centerX = (int)((landmarks.UpperLipTop.X + landmarks.UnderLipBottom.X) / 2); int centerY = (int)((landmarks.UpperLipTop.Y + landmarks.UnderLipBottom.Y) / 2); var mouthRect = new Rect(centerX - boxWidth / 2, centerY - boxHeight / 2, boxWidth, boxHeight); // Draw Mouth Box (Blue in BGR) Cv2.Rectangle(frame, mouthRect, new Scalar(255, 0, 0), 2); // Add Label Cv2.PutText(frame, "Mouth", new Point(mouthRect.X, mouthRect.Y - 5), HersheyFonts.HersheySimplex, 0.4, new Scalar(255, 0, 0), 1); } 4.3 Face Bounding Box /// <summary> /// Draws rectangle around entire face using the face rectangle from API. /// </summary> public void DrawFaceBox(FaceDetectionResult face, Mat frame) { var faceRect = face.FaceRectangle; if (faceRect == null) { return; } var rect = new Rect( faceRect.Left, faceRect.Top, faceRect.Width, faceRect.Height ); // Draw Face Bounding Box (Red in BGR) Cv2.Rectangle(frame, rect, new Scalar(0, 0, 255), 2); // Add Label with dimensions Cv2.PutText(frame, $"Face {faceRect.Width}x{faceRect.Height}", new Point(rect.X, rect.Y - 10), HersheyFonts.HersheySimplex, 0.5, new Scalar(0, 0, 255), 2); } 4.4 Nose Bounding Box /// <summary> /// Draws bounding box around nose using nose landmarks. /// </summary> public void DrawNoseBox(FaceLandmarks landmarks, Mat frame) { // Calculate horizontal bounds from Alar tips int minX = (int)Math.Min(landmarks.NoseLeftAlarOutTip.X, landmarks.NoseRightAlarOutTip.X); int maxX = (int)Math.Max(landmarks.NoseLeftAlarOutTip.X, landmarks.NoseRightAlarOutTip.X); // Calculate vertical bounds from Root to Tip int minY = (int)Math.Min(landmarks.NoseRootLeft.Y, landmarks.NoseTip.Y); int maxY = (int)landmarks.NoseTip.Y; // Create Rect with a 10px padding buffer var noseRect = new Rect( minX - 10, minY - 10, (maxX - minX) + 20, (maxY - minY) + 20 ); // Draw Nose Box (Yellow in BGR) Cv2.Rectangle(frame, noseRect, new Scalar(0, 255, 255), 2); } Part 5: Geometric Calculations with Landmarks 5.1 Calculating Euclidean Distance /// <summary> /// Calculates distance between two landmark points. /// </summary> public static double CalculateDistance(dynamic point1, dynamic point2) { double dx = point1.X - point2.X; double dy = point1.Y - point2.Y; return Math.Sqrt(dx * dx + dy * dy); } 5.2 Eye Aspect Ratio (EAR) Formula /// <summary> /// Calculates the Eye Aspect Ratio (EAR) to detect eye closure. /// </summary> public double CalculateEAR( FaceLandmarkCoordinate top1, FaceLandmarkCoordinate top2, FaceLandmarkCoordinate bottom1, FaceLandmarkCoordinate bottom2, FaceLandmarkCoordinate inner, FaceLandmarkCoordinate outer) { // Vertical distances double v1 = CalculateDistance(top1, bottom1); double v2 = CalculateDistance(top2, bottom2); // Horizontal distance double h = CalculateDistance(inner, outer); // EAR formula: (||p2-p6|| + ||p3-p5||) / (2 * ||p1-p4||) return (v1 + v2) / (2.0 * h); } Simplified Implementation: /// <summary> /// Calculates Eye Aspect Ratio (EAR) for a single eye. /// Reference: "Real-Time Eye Blink Detection using Facial Landmarks" (Soukupová & Čech, 2016) /// </summary> public double ComputeEAR(FaceLandmarks landmarks, bool isLeftEye) { var top = isLeftEye ? landmarks.EyeLeftTop : landmarks.EyeRightTop; var bottom = isLeftEye ? landmarks.EyeLeftBottom : landmarks.EyeRightBottom; var inner = isLeftEye ? landmarks.EyeLeftInner : landmarks.EyeRightInner; var outer = isLeftEye ? landmarks.EyeLeftOuter : landmarks.EyeRightOuter; if (top == null || bottom == null || inner == null || outer == null) { _logger.LogWarning("Missing eye landmarks"); return 1.0; // Return 1.0 (open) to prevent false positives for drowsiness } double verticalDist = CalculateDistance(top, bottom); double horizontalDist = CalculateDistance(inner, outer); // Simplified EAR for Azure 27-point model double ear = verticalDist / horizontalDist; _logger.LogDebug( "EAR for {Eye}: {Value:F3}", isLeftEye ? "left" : "right", ear ); return ear; } Usage Example: var leftEAR = ComputeEAR(landmarks, isLeftEye: true); var rightEAR = ComputeEAR(landmarks, isLeftEye: false); var avgEAR = (leftEAR + rightEAR) / 2.0; Console.WriteLine($"Average EAR: {avgEAR:F3}"); // Open eyes: ~0.25-0.30 // Closed eyes: ~0.10-0.15 5.3 Mouth Aspect Ratio (MAR) /// <summary> /// Calculates Mouth Aspect Ratio relative to face height. /// </summary> public double CalculateMouthAspectRatio(FaceLandmarks landmarks, FaceRectangle faceRect) { double mouthHeight = landmarks.UnderLipBottom.Y - landmarks.UpperLipTop.Y; double mouthWidth = CalculateDistance(landmarks.MouthLeft, landmarks.MouthRight); double mouthOpenRatio = mouthHeight / faceRect.Height; double mouthWidthRatio = mouthWidth / faceRect.Width; _logger.LogDebug( "Mouth - Height ratio: {HeightRatio:F3}, Width ratio: {WidthRatio:F3}", mouthOpenRatio, mouthWidthRatio ); return mouthOpenRatio; } 5.4 Inter-Eye Distance /// <summary> /// Calculates the distance between pupils (inter-pupillary distance). /// </summary> public double CalculateInterEyeDistance(FaceLandmarks landmarks) { return CalculateDistance(landmarks.PupilLeft, landmarks.PupilRight); } /// <summary> /// Calculates distance between inner eye corners. /// </summary> public double CalculateInnerEyeDistance(FaceLandmarks landmarks) { return CalculateDistance(landmarks.EyeLeftInner, landmarks.EyeRightInner); } 5.5 Face Symmetry Analysis /// <summary> /// Analyzes facial symmetry by comparing left and right sides. /// </summary> public FaceSymmetryMetrics AnalyzeFaceSymmetry(FaceLandmarks landmarks) { double centerX = landmarks.NoseTip.X; double leftEyeDistance = CalculateDistance(landmarks.EyeLeftInner, new { X = centerX, Y = landmarks.EyeLeftInner.Y }); double leftMouthDistance = CalculateDistance(landmarks.MouthLeft, new { X = centerX, Y = landmarks.MouthLeft.Y }); double rightEyeDistance = CalculateDistance(landmarks.EyeRightInner, new { X = centerX, Y = landmarks.EyeRightInner.Y }); double rightMouthDistance = CalculateDistance(landmarks.MouthRight, new { X = centerX, Y = landmarks.MouthRight.Y }); return new FaceSymmetryMetrics { EyeSymmetryRatio = leftEyeDistance / rightEyeDistance, MouthSymmetryRatio = leftMouthDistance / rightMouthDistance, IsSymmetric = Math.Abs(leftEyeDistance - rightEyeDistance) < 5.0 }; } public class FaceSymmetryMetrics { public double EyeSymmetryRatio { get; set; } public double MouthSymmetryRatio { get; set; } public bool IsSymmetric { get; set; } } Part 6: Head Pose Estimation 6.1 Understanding Head Pose Angles Azure Face API provides three Euler angles for head orientation: 6.2 Accessing Head Pose Data public void AnalyzeHeadPose(FaceDetectionResult face) { var headPose = face.FaceAttributes?.HeadPose; if (headPose == null) { _logger.LogWarning("Head pose not available"); return; } double yaw = headPose.Yaw; double pitch = headPose.Pitch; double roll = headPose.Roll; Console.WriteLine("Head Pose:"); Console.WriteLine($" Yaw: {yaw:F2}° (Left/Right)"); Console.WriteLine($" Pitch: {pitch:F2}° (Up/Down)"); Console.WriteLine($" Roll: {roll:F2}° (Tilt)"); InterpretHeadPose(yaw, pitch, roll); } 6.3 Interpreting Head Pose public string InterpretHeadPose(double yaw, double pitch, double roll) { var directions = new List<string>(); // Interpret Yaw (horizontal) if (Math.Abs(yaw) < 10) directions.Add("Looking Forward"); else if (yaw < -20) directions.Add($"Turned Left ({Math.Abs(yaw):F0}°)"); else if (yaw > 20) directions.Add($"Turned Right ({yaw:F0}°)"); // Interpret Pitch (vertical) if (Math.Abs(pitch) < 10) directions.Add("Level"); else if (pitch < -15) directions.Add($"Looking Down ({Math.Abs(pitch):F0}°)"); else if (pitch > 15) directions.Add($"Looking Up ({pitch:F0}°)"); // Interpret Roll (tilt) if (Math.Abs(roll) > 15) { string side = roll < 0 ? "Left" : "Right"; directions.Add($"Tilted {side} ({Math.Abs(roll):F0}°)"); } return string.Join(", ", directions); } 6.4 Visualizing Head Pose on Frame /// <summary> /// Draws head pose information with color-coded indicators. /// </summary> public void DrawHeadPoseInfo(Mat frame, HeadPose headPose, FaceRectangle faceRect) { double yaw = headPose.Yaw; double pitch = headPose.Pitch; double roll = headPose.Roll; int centerX = faceRect.Left + faceRect.Width / 2; int centerY = faceRect.Top + faceRect.Height / 2; string poseText = $"Yaw: {yaw:F1}° Pitch: {pitch:F1}° Roll: {roll:F1}°"; Cv2.PutText(frame, poseText, new Point(faceRect.Left, faceRect.Top - 10), HersheyFonts.HersheySimplex, 0.5, new Scalar(255, 255, 255), 1); int arrowLength = 50; double yawRadians = yaw * Math.PI / 180.0; int arrowEndX = centerX + (int)(arrowLength * Math.Sin(yawRadians)); Cv2.ArrowedLine(frame, new Point(centerX, centerY), new Point(arrowEndX, centerY), new Scalar(0, 255, 0), 2, tipLength: 0.3); double pitchRadians = -pitch * Math.PI / 180.0; int arrowPitchEndY = centerY + (int)(arrowLength * Math.Sin(pitchRadians)); Cv2.ArrowedLine(frame, new Point(centerX, centerY), new Point(centerX, arrowPitchEndY), new Scalar(255, 0, 0), 2, tipLength: 0.3); } 6.5 Detecting Head Orientation States public enum HeadOrientation { Forward, Left, Right, Up, Down, TiltedLeft, TiltedRight, UpLeft, UpRight, DownLeft, DownRight } public List<HeadOrientation> DetectHeadOrientation(HeadPose headPose) { const double THRESHOLD = 15.0; bool lookingUp = headPose.Pitch > THRESHOLD; bool lookingDown = headPose.Pitch < -THRESHOLD; bool lookingLeft = headPose.Yaw < -THRESHOLD; bool lookingRight = headPose.Yaw > THRESHOLD; var orientations = new List<HeadOrientation>(); if (!lookingUp && !lookingDown && !lookingLeft && !lookingRight) orientations.Add(HeadOrientation.Forward); if (lookingUp && !lookingLeft && !lookingRight) orientations.Add(HeadOrientation.Up); if (lookingDown && !lookingLeft && !lookingRight) orientations.Add(HeadOrientation.Down); if (lookingLeft && !lookingUp && !lookingDown) orientations.Add(HeadOrientation.Left); if (lookingRight && !lookingUp && !lookingDown) orientations.Add(HeadOrientation.Right); if (lookingUp && lookingLeft) orientations.Add(HeadOrientation.UpLeft); if (lookingUp && lookingRight) orientations.Add(HeadOrientation.UpRight); if (lookingDown && lookingLeft) orientations.Add(HeadOrientation.DownLeft); if (lookingDown && lookingRight) orientations.Add(HeadOrientation.DownRight); return orientations; } Part 7: Real-Time Video Processing 7.1 Setting Up Video Capture using OpenCvSharp; public class RealTimeFaceAnalyzer : IDisposable { private VideoCapture? _capture; private Mat? _frame; private readonly FaceClient _faceClient; private bool _isRunning; public async Task StartAsync() { _capture = new VideoCapture(0); _frame = new Mat(); _isRunning = true; await Task.Run(() => ProcessVideoLoop()); } private async Task ProcessVideoLoop() { while (_isRunning) { if (_capture == null || !_capture.IsOpened()) break; _capture.Read(_frame); if (_frame == null || _frame.Empty()) { await Task.Delay(1); // Minimal delay to prevent CPU spiking continue; } Cv2.Resize(_frame, _frame, new Size(640, 480)); // Ensure we don't await indefinitely in the rendering loop _ = ProcessFrameAsync(_frame.Clone()); Cv2.ImShow("Face Analysis", _frame); if (Cv2.WaitKey(30) == 'q') break; } Dispose(); } private async Task ProcessFrameAsync(Mat frame) { // This is where your DrawFaceBox, DrawAllLandmarks, and EAR logic will sit. // Remember to use try-catch here to prevent API errors from crashing the loop. } public void Dispose() { _isRunning = false; _capture?.Dispose(); _frame?.Dispose(); Cv2.DestroyAllWindows(); } } 7.2 Optimizing API Calls Problem: Calling Azure Face API on every frame (30 fps) is expensive and slow. Solution: Call API once per second, cache results for 30 frames. private List<FaceDetectionResult> _cachedFaces = new(); private DateTime _lastDetectionTime = DateTime.MinValue; private readonly object _cacheLock = new(); private async Task ProcessFrameAsync(Mat frame) { if ((DateTime.Now - _lastDetectionTime).TotalSeconds >= 1.0) { _lastDetectionTime = DateTime.Now; byte[] imageBytes; Cv2.ImEncode(".jpg", frame, out imageBytes); var faces = await DetectFacesAsync(imageBytes); lock (_cacheLock) { _cachedFaces = faces; } } List<FaceDetectionResult> facesToProcess; lock (_cacheLock) { facesToProcess = _cachedFaces.ToList(); } foreach (var face in facesToProcess) { DrawFaceAnnotations(face, frame); } } Performance Improvement: 30x fewer API calls (1/sec instead of 30/sec) ~$0.02/hour instead of ~$0.60/hour Smooth 30 fps rendering < 100ms latency for visual updates 7.3 Drawing Complete Face Annotations private void DrawFaceAnnotations(FaceDetectionResult face, Mat frame) { DrawFaceBox(face, frame); if (face.FaceLandmarks != null) { DrawAllLandmarks(face.FaceLandmarks, frame); DrawEyeBoxes(face.FaceLandmarks, frame); DrawMouthBox(face.FaceLandmarks, frame); DrawNoseBox(face.FaceLandmarks, frame); double leftEAR = ComputeEAR(face.FaceLandmarks, isLeftEye: true); double rightEAR = ComputeEAR(face.FaceLandmarks, isLeftEye: false); double avgEAR = (leftEAR + rightEAR) / 2.0; Cv2.PutText(frame, $"EAR: {avgEAR:F3}", new Point(10, 30), HersheyFonts.HersheySimplex, 0.6, new Scalar(0, 255, 0), 2); } if (face.FaceAttributes?.HeadPose != null) { DrawHeadPoseInfo(frame, face.FaceAttributes.HeadPose, face.FaceRectangle); string orientation = InterpretHeadPose(face.FaceAttributes.HeadPose.Yaw, face.FaceAttributes.HeadPose.Pitch, face.FaceAttributes.HeadPose.Roll); Cv2.PutText(frame, orientation, new Point(10, 60), HersheyFonts.HersheySimplex, 0.6, new Scalar(255, 255, 0), 2); } } Part 8: Advanced Features and Use Cases 8.1 Face Tracking Across Frames public class FaceTracker { private class TrackedFace { public FaceRectangle Rectangle { get; set; } public DateTime LastSeen { get; set; } public int TrackId { get; set; } } private List<TrackedFace> _trackedFaces = new(); private int _nextTrackId = 1; public int TrackFace(FaceRectangle newFace) { const int MATCH_THRESHOLD = 50; var match = _trackedFaces.FirstOrDefault(tf => { double distance = Math.Sqrt(Math.Pow(tf.Rectangle.Left - newFace.Left, 2) + Math.Pow(tf.Rectangle.Top - newFace.Top, 2)); return distance < MATCH_THRESHOLD; }); if (match != null) { match.Rectangle = newFace; match.LastSeen = DateTime.Now; return match.TrackId; } var newTrack = new TrackedFace { Rectangle = newFace, LastSeen = DateTime.Now, TrackId = _nextTrackId++ }; _trackedFaces.Add(newTrack); return newTrack.TrackId; } public void RemoveOldTracks(TimeSpan maxAge) { _trackedFaces.RemoveAll(tf => DateTime.Now - tf.LastSeen > maxAge); } } 8.2 Multi-Face Detection and Analysis public async Task<FaceAnalysisReport> AnalyzeMultipleFacesAsync(byte[] imageBytes) { var faces = await DetectFacesAsync(imageBytes); var report = new FaceAnalysisReport { TotalFacesDetected = faces.Count, Timestamp = DateTime.Now, Faces = new List<SingleFaceAnalysis>() }; for (int i = 0; i < faces.Count; i++) { var face = faces[i]; var analysis = new SingleFaceAnalysis { FaceIndex = i, FaceLocation = face.FaceRectangle, FaceSize = face.FaceRectangle.Width * face.FaceRectangle.Height }; if (face.FaceLandmarks != null) { analysis.LeftEyeEAR = ComputeEAR(face.FaceLandmarks, true); analysis.RightEyeEAR = ComputeEAR(face.FaceLandmarks, false); analysis.InterPupillaryDistance = CalculateInterEyeDistance(face.FaceLandmarks); } if (face.FaceAttributes?.HeadPose != null) { analysis.HeadYaw = face.FaceAttributes.HeadPose.Yaw; analysis.HeadPitch = face.FaceAttributes.HeadPose.Pitch; analysis.HeadRoll = face.FaceAttributes.HeadPose.Roll; } report.Faces.Add(analysis); } report.Faces = report.Faces.OrderByDescending(f => f.FaceSize).ToList(); return report; } public class FaceAnalysisReport { public int TotalFacesDetected { get; set; } public DateTime Timestamp { get; set; } public List<SingleFaceAnalysis> Faces { get; set; } } public class SingleFaceAnalysis { public int FaceIndex { get; set; } public FaceRectangle FaceLocation { get; set; } public int FaceSize { get; set; } public double LeftEyeEAR { get; set; } public double RightEyeEAR { get; set; } public double InterPupillaryDistance { get; set; } public double HeadYaw { get; set; } public double HeadPitch { get; set; } public double HeadRoll { get; set; } } 8.3 Exporting Landmark Data to JSON using System.Text.Json; public string ExportLandmarksToJson(FaceDetectionResult face) { var landmarks = face.FaceLandmarks; var landmarkData = new { Face = new { Rectangle = new { face.FaceRectangle.Left, face.FaceRectangle.Top, face.FaceRectangle.Width, face.FaceRectangle.Height } }, Eyes = new { Left = new { Outer = new { landmarks.EyeLeftOuter.X, landmarks.EyeLeftOuter.Y }, Inner = new { landmarks.EyeLeftInner.X, landmarks.EyeLeftInner.Y }, Top = new { landmarks.EyeLeftTop.X, landmarks.EyeLeftTop.Y }, Bottom = new { landmarks.EyeLeftBottom.X, landmarks.EyeLeftBottom.Y } }, Right = new { Outer = new { landmarks.EyeRightOuter.X, landmarks.EyeRightOuter.Y }, Inner = new { landmarks.EyeRightInner.X, landmarks.EyeRightInner.Y }, Top = new { landmarks.EyeRightTop.X, landmarks.EyeRightTop.Y }, Bottom = new { landmarks.EyeRightBottom.X, landmarks.EyeRightBottom.Y } } }, Mouth = new { UpperLipTop = new { landmarks.UpperLipTop.X, landmarks.UpperLipTop.Y }, UnderLipBottom = new { landmarks.UnderLipBottom.X, landmarks.UnderLipBottom.Y }, Left = new { landmarks.MouthLeft.X, landmarks.MouthLeft.Y }, Right = new { landmarks.MouthRight.X, landmarks.MouthRight.Y } }, Nose = new { Tip = new { landmarks.NoseTip.X, landmarks.NoseTip.Y }, RootLeft = new { landmarks.NoseRootLeft.X, landmarks.NoseRootLeft.Y }, RootRight = new { landmarks.NoseRootRight.X, landmarks.NoseRootRight.Y } }, HeadPose = face.FaceAttributes?.HeadPose != null ? new { face.FaceAttributes.HeadPose.Yaw, face.FaceAttributes.HeadPose.Pitch, face.FaceAttributes.HeadPose.Roll } : null }; return JsonSerializer.Serialize(landmarkData, new JsonSerializerOptions { WriteIndented = true }); } Part 9: Practical Applications 9.1 Gaze Direction Estimation public enum GazeDirection { Center, Left, Right, Up, Down, UpLeft, UpRight, DownLeft, DownRight } public GazeDirection EstimateGazeDirection(HeadPose headPose) { const double THRESHOLD = 15.0; bool lookingUp = headPose.Pitch > THRESHOLD; bool lookingDown = headPose.Pitch < -THRESHOLD; bool lookingLeft = headPose.Yaw < -THRESHOLD; bool lookingRight = headPose.Yaw > THRESHOLD; if (lookingUp && lookingLeft) return GazeDirection.UpLeft; if (lookingUp && lookingRight) return GazeDirection.UpRight; if (lookingDown && lookingLeft) return GazeDirection.DownLeft; if (lookingDown && lookingRight) return GazeDirection.DownRight; if (lookingUp) return GazeDirection.Up; if (lookingDown) return GazeDirection.Down; if (lookingLeft) return GazeDirection.Left; if (lookingRight) return GazeDirection.Right; return GazeDirection.Center; } 9.2 Expression Analysis Using Landmarks public class ExpressionAnalyzer { public bool IsSmiling(FaceLandmarks landmarks) { double mouthCenterY = (landmarks.UpperLipTop.Y + landmarks.UnderLipBottom.Y) / 2; double leftCornerY = landmarks.MouthLeft.Y; double rightCornerY = landmarks.MouthRight.Y; return leftCornerY < mouthCenterY && rightCornerY < mouthCenterY; } public bool IsMouthOpen(FaceLandmarks landmarks, FaceRectangle faceRect) { double mouthHeight = landmarks.UnderLipBottom.Y - landmarks.UpperLipTop.Y; double mouthOpenRatio = mouthHeight / faceRect.Height; return mouthOpenRatio > 0.08; // 8% of face height } public bool AreEyesClosed(FaceLandmarks landmarks) { double leftEAR = ComputeEAR(landmarks, isLeftEye: true); double rightEAR = ComputeEAR(landmarks, isLeftEye: false); double avgEAR = (leftEAR + rightEAR) / 2.0; return avgEAR < 0.18; // Threshold for closed eyes } } 9.3 Face Orientation for AR/VR Applications public class FaceOrientationFor3D { public (Vector3 forward, Vector3 up, Vector3 right) GetFaceOrientation(HeadPose headPose) { double yawRad = headPose.Yaw * Math.PI / 180.0; double pitchRad = headPose.Pitch * Math.PI / 180.0; double rollRad = headPose.Roll * Math.PI / 180.0; var forward = new Vector3((float)(Math.Sin(yawRad) * Math.Cos(pitchRad)), (float)(-Math.Sin(pitchRad)), (float)(Math.Cos(yawRad) * Math.Cos(pitchRad))); var up = new Vector3((float)(Math.Sin(yawRad) * Math.Sin(pitchRad) * Math.Cos(rollRad) - Math.Cos(yawRad) * Math.Sin(rollRad)), (float)(Math.Cos(pitchRad) * Math.Cos(rollRad)), (float)(Math.Cos(yawRad) * Math.Sin(pitchRad) * Math.Cos(rollRad) + Math.Sin(yawRad) * Math.Sin(rollRad))); var right = Vector3.Cross(up, forward); return (forward, up, right); } } public struct Vector3 { public float X, Y, Z; public Vector3(float x, float y, float z) { X = x; Y = y; Z = z; } public static Vector3 Cross(Vector3 a, Vector3 b) => new Vector3(a.Y * b.Z - a.Z * b.Y, a.Z * b.X - a.X * b.Z, a.X * b.Y - a.Y * b.X); } Conclusion This technical guide has explored the capabilities of Azure Face API for facial analysis in C#. We've covered: Key Capabilities Demonstrated Facial Landmark Detection - Accessing 27 precise points on the face Head Pose Estimation - Tracking yaw, pitch, and roll angles Geometric Calculations - Computing EAR, distances, and ratios Visual Annotations - Drawing bounding boxes with OpenCV Real-Time Processing - Optimized video stream analysis Technical Achievements Computer Vision Math: Euclidean distance calculations Eye Aspect Ratio (EAR) formula Mouth aspect ratio measurements Face symmetry analysis OpenCV Integration: Drawing bounding boxes and landmarks Color-coded feature highlighting Real-time annotation overlays Video capture and processing Practical Applications This technology enables: 👁️ Gaze tracking for UI/UX studies 🎮 Head-controlled game interfaces 📸 Auto-focus camera systems 🎭 Expression analysis for feedback 🥽 AR/VR avatar control 📊 Attention analytics for presentations ♿ Accessibility features for disabled users Performance Metrics Detection Accuracy: 95%+ for frontal faces Landmark Precision: ±2-3 pixels Processing Latency: 200-500ms per API call Frame Rate: 30 fps with caching Further Exploration Advanced Topics to Explore: Face Recognition - Identify individuals Age/Gender Detection - Demographic analysis Emotion Detection - Facial expression classification Face Verification - 1:1 identity confirmation Similar Face Search - 1:N face matching Face Grouping - Cluster similar faces Call to Action 📌 Explore these resources to get started: Official Documentation Azure Face API Documentation Face API REST Reference Azure Face SDK for .NET Related Libraries OpenCVSharp - OpenCV wrapper for .NET System.Drawing - .NET image processing Source Code GitHub Repository: ravimodi_microsoft/SmartDriver Sample Code: Included in this articleBuilding HIPAA-Compliant Medical Transcription with Local AI
Building HIPAA-Compliant Medical Transcription with Local AI Introduction Healthcare organizations generate vast amounts of spoken content, patient consultations, research interviews, clinical notes, medical conferences. Transcribing these recordings traditionally requires either manual typing (time-consuming and expensive) or cloud transcription services (creating immediate HIPAA compliance concerns). Every audio file sent to external APIs exposes Protected Health Information (PHI), requires Business Associate Agreements, creates audit trails on third-party servers, and introduces potential breach vectors. This sample solution lies in on-premises voice-to-text systems that process audio entirely locally, never sending PHI beyond organizational boundaries. This article demonstrates building a sample medical transcription application using FLWhisper, ASP.NET Core, C#, and Microsoft Foundry Local with OpenAI Whisper models. You'll learn how to build sample HIPAA-compliant audio processing, integrate Whisper models for medical terminology accuracy, design privacy-first API patterns, and build responsive web UIs for healthcare workflows. Whether you're developing electronic health record (EHR) integrations, building clinical research platforms, or implementing dictation systems for medical practices, this sample could be a great starting point for privacy-first speech recognition. Why Local Transcription Is Critical for Healthcare Healthcare data handling is fundamentally different from general business data due to HIPAA regulations, state privacy laws, and professional ethics obligations. Understanding these requirements explains why cloud transcription services, despite their convenience, create unacceptable risks for medical applications. HIPAA compliance mandates strict controls over PHI. Every system that touches patient data must implement administrative, physical, and technical safeguards. Cloud transcription APIs require Business Associate Agreements (BAAs), but even with paperwork, you're entrusting PHI to external systems. Every API call creates logs on vendor servers, potentially in multiple jurisdictions. Data breaches at transcription vendors expose patient information, creating liability for healthcare organizations. On-premises processing eliminates these third-party risks entirely, PHI never leaves your controlled environment. US State laws increasingly add requirements beyond HIPAA. California's CCPA, New York's SHIELD Act, and similar legislation create additional compliance obligations. International regulations like GDPR prohibit transferring health data outside approved jurisdictions. Local processing simplifies compliance by keeping data within organizational boundaries. Research applications face even stricter requirements. Institutional Review Boards (IRBs) often require explicit consent for data sharing with external parties. Cloud transcription may violate study protocols that promise "no third-party data sharing." Clinical trials in pharmaceutical development handle proprietary information alongside PHI, double jeopardy for data exposure. Local transcription maintains research integrity while enabling audio analysis. Cost considerations favor local deployment at scale. Medical organizations generate substantial audio, thousands of patient encounters monthly. Cloud APIs charge per minute of audio, creating significant recurring costs. Local models have fixed infrastructure costs that scale economically. A modest GPU server can process hundreds of hours monthly at predictable expense. Latency matters for clinical workflows. Doctors and nurses need transcriptions available immediately after patient encounters to review and edit while details are fresh. Cloud APIs introduce network delays, especially problematic in rural health facilities with limited connectivity. Local inference provides <1 second turnaround for typical consultation lengths. Application Architecture: ASP.NET Core with Foundry Local The sample FLWhisper application implements clean separation between audio handling, AI inference, and state management using modern .NET patterns: The ASP.NET Core 10 minimal API provides HTTP endpoints for health checks, audio transcription, and sample file streaming. Minimal APIs reduce boilerplate while maintaining full middleware support for error handling, authentication, and CORS. The API design follows OpenAI's transcription endpoint specification, enabling drop-in replacement for existing integrations. The service layer encapsulates business logic: FoundryModelService manages model loading and lifetime, TranscriptionService handles audio processing and AI inference, and SampleAudioService provides demonstration files for testing. This separation enables easy testing, dependency injection, and service swapping. Foundry Local integration uses the Microsoft.AI.Foundry.Local.WinML SDK. Unlike cloud APIs requiring authentication and network calls, this SDK communicates directly with the local Foundry service via in-process calls. Models load once at startup, remaining resident in memory for sub-second inference on subsequent requests. The static file frontend delivers vanilla HTML/CSS/JavaScript, no framework overhead. This simplicity aids healthcare IT security audits and enables deployment on locked-down hospital networks. The UI provides file upload, sample selection, audio preview, transcription requests, and result display with copy-to-clipboard functionality. Here's the architectural flow for transcription requests: Web UI (Upload Audio File) ↓ POST /v1/audio/transcriptions (Multipart Form Data) ↓ ASP.NET Core API Route ↓ TranscriptionService.TranscribeAudio(audioStream) ↓ Foundry Local Model (Whisper Medium locally) ↓ Text Result + Metadata (language, duration) ↓ Return JSON/Text Response ↓ Display in UI This architecture embodies several healthcare system design principles: Data never leaves the device: All processing occurs on-premises, no external API calls No data persistence by default: Audio and transcripts are session-only, never saved unless explicitly configured Comprehensive health checks: System readiness verification before accepting PHI Audit logging support: Structured logging for compliance documentation Graceful degradation: Clear error messages when models unavailable rather than silent failures Setting Up Foundry Local with Whisper Models Foundry Local supports multiple Whisper model sizes, each with different accuracy/speed tradeoffs. For medical transcription, accuracy is paramount—misheard drug names or dosages create patient safety risks: # Install Foundry Local (Windows) winget install Microsoft.FoundryLocal # Verify installation foundry --version # Download Whisper Medium model (optimal for medical accuracy) foundry model add openai-whisper-medium-generic-cpu:1 # Check model availability foundry model list Whisper Medium (769M parameters) provides the best balance for medical use. Smaller models (Tiny, Base) miss medical terminology frequently. Larger models (Large) offer marginal accuracy gains at 3x inference time. Medium handles medical vocabulary well, drug names, anatomical terms, procedure names, while processing typical consultation audio (5-10 minutes) in under 30 seconds. The application detects and loads the model automatically: // Services/FoundryModelService.cs using Microsoft.AI.Foundry.Local.WinML; public class FoundryModelService { private readonly ILogger _logger; private readonly FoundryOptions _options; private ILocalAIModel? _loadedModel; public FoundryModelService( ILogger logger, IOptions options) { _logger = logger; _options = options.Value; } public async Task InitializeModelAsync() { try { _logger.LogInformation( "Loading Foundry model: {ModelAlias}", _options.ModelAlias ); // Load model from Foundry Local _loadedModel = await FoundryClient.LoadModelAsync( modelAlias: _options.ModelAlias, cancellationToken: CancellationToken.None ); if (_loadedModel == null) { _logger.LogWarning("Model loaded but returned null instance"); return false; } _logger.LogInformation( "Successfully loaded model: {ModelAlias}", _options.ModelAlias ); return true; } catch (Exception ex) { _logger.LogError( ex, "Failed to load Foundry model: {ModelAlias}", _options.ModelAlias ); return false; } } public ILocalAIModel? GetLoadedModel() => _loadedModel; public async Task UnloadModelAsync() { if (_loadedModel != null) { await FoundryClient.UnloadModelAsync(_loadedModel); _loadedModel = null; _logger.LogInformation("Model unloaded"); } } } Configuration lives in appsettings.json , enabling easy customization without code changes: { "Foundry": { "ModelAlias": "whisper-medium", "LogLevel": "Information" }, "Transcription": { "MaxAudioDurationSeconds": 300, "SupportedFormats": ["wav", "mp3", "m4a", "flac"], "DefaultLanguage": "en" } } Implementing Privacy-First Transcription Service The transcription service handles audio processing while maintaining strict privacy controls. No audio or transcript persists beyond the HTTP request lifecycle unless explicitly configured: // Services/TranscriptionService.cs public class TranscriptionService { private readonly FoundryModelService _modelService; private readonly ILogger _logger; public async Task TranscribeAudioAsync( Stream audioStream, string originalFileName, TranscriptionOptions? options = null) { options ??= new TranscriptionOptions(); var startTime = DateTime.UtcNow; try { // Validate audio format ValidateAudioFormat(originalFileName); // Get loaded model var model = _modelService.GetLoadedModel(); if (model == null) { throw new InvalidOperationException("Whisper model not loaded"); } // Create temporary file (automatically deleted after transcription) using var tempFile = new TempAudioFile(audioStream); // Execute transcription _logger.LogInformation( "Starting transcription for file: {FileName}", originalFileName ); var transcription = await model.TranscribeAsync( audioFilePath: tempFile.Path, language: options.Language, cancellationToken: CancellationToken.None ); var duration = (DateTime.UtcNow - startTime).TotalSeconds; _logger.LogInformation( "Transcription completed in {Duration:F2}s", duration ); return new TranscriptionResult { Text = transcription.Text, Language = transcription.Language ?? options.Language, Duration = transcription.AudioDuration, ProcessingTimeSeconds = duration, FileName = originalFileName, Timestamp = DateTime.UtcNow }; } catch (Exception ex) { _logger.LogError( ex, "Transcription failed for file: {FileName}", originalFileName ); throw; } } private void ValidateAudioFormat(string fileName) { var extension = Path.GetExtension(fileName).TrimStart('.'); var supportedFormats = new[] { "wav", "mp3", "m4a", "flac", "ogg" }; if (!supportedFormats.Contains(extension.ToLowerInvariant())) { throw new ArgumentException( $"Unsupported audio format: {extension}. " + $"Supported: {string.Join(", ", supportedFormats)}" ); } } } // Temporary file wrapper that auto-deletes internal class TempAudioFile : IDisposable { public string Path { get; } public TempAudioFile(Stream sourceStream) { Path = System.IO.Path.GetTempFileName(); using var fileStream = File.OpenWrite(Path); sourceStream.CopyTo(fileStream); } public void Dispose() { try { if (File.Exists(Path)) { File.Delete(Path); } } catch { // Ignore deletion errors in temp folder } } } This service demonstrates several privacy-first patterns: Temporary file lifecycle management: Audio written to temp storage, automatically deleted after transcription No implicit persistence: Results returned to caller, not saved by service Format validation: Accept only supported audio formats to prevent processing errors Comprehensive logging: Audit trail for compliance without logging PHI content Error isolation: Exceptions contain diagnostic info but no patient data Building the OpenAI-Compatible REST API The API endpoint mirrors OpenAI's transcription API specification, enabling existing integrations to work without modifications: // Program.cs var builder = WebApplication.CreateBuilder(args); // Configure services builder.Services.Configure( builder.Configuration.GetSection("Foundry") ); builder.Services.AddSingleton(); builder.Services.AddScoped(); builder.Services.AddHealthChecks() .AddCheck("foundry-health"); var app = builder.Build(); // Load model at startup var modelService = app.Services.GetRequiredService(); await modelService.InitializeModelAsync(); app.UseHealthChecks("/health"); app.MapHealthChecks("/api/health/status"); // OpenAI-compatible transcription endpoint app.MapPost("/v1/audio/transcriptions", async ( HttpRequest request, TranscriptionService transcriptionService, ILogger logger) => { if (!request.HasFormContentType) { return Results.BadRequest(new { error = "Content-Type must be multipart/form-data" }); } var form = await request.ReadFormAsync(); // Extract audio file var audioFile = form.Files.GetFile("file"); if (audioFile == null || audioFile.Length == 0) { return Results.BadRequest(new { error = "Audio file required in 'file' field" }); } // Parse options var format = form["format"].ToString() ?? "text"; var language = form["language"].ToString() ?? "en"; try { // Process transcription using var stream = audioFile.OpenReadStream(); var result = await transcriptionService.TranscribeAudioAsync( audioStream: stream, originalFileName: audioFile.FileName, options: new TranscriptionOptions { Language = language } ); // Return in requested format if (format == "json") { return Results.Json(new { text = result.Text, language = result.Language, duration = result.Duration }); } else { // Default: plain text return Results.Text(result.Text); } } catch (Exception ex) { logger.LogError(ex, "Transcription request failed"); return Results.StatusCode(500); } }) .DisableAntiforgery() // File uploads need CSRF exemption .WithName("TranscribeAudio") .WithOpenApi(); app.Run(); Example API usage: # PowerShell $audioFile = Get-Item "consultation-recording.wav" $response = Invoke-RestMethod ` -Uri "http://localhost:5192/v1/audio/transcriptions" ` -Method Post ` -Form @{ file = $audioFile; format = "json" } Write-Output $response.text # cURL curl -X POST http://localhost:5192/v1/audio/transcriptions \ -F "file=@consultation-recording.wav" \ -F "format=json" Building the Interactive Web Frontend The web UI provides a user-friendly interface for non-technical medical staff to transcribe recordings: SarahCare Medical Transcription The JavaScript handles file uploads and API interactions: // wwwroot/app.js let selectedFile = null; async function checkHealth() { try { const response = await fetch('/health'); const statusEl = document.getElementById('status'); if (response.ok) { statusEl.className = 'status-badge online'; statusEl.textContent = '✓ System Ready'; } else { statusEl.className = 'status-badge offline'; statusEl.textContent = '✗ System Unavailable'; } } catch (error) { console.error('Health check failed:', error); } } function handleFileSelect(event) { const file = event.target.files[0]; if (!file) return; selectedFile = file; // Show file info const fileInfo = document.getElementById('fileInfo'); fileInfo.textContent = `Selected: ${file.name} (${formatFileSize(file.size)})`; fileInfo.classList.remove('hidden'); // Enable audio preview const preview = document.getElementById('audioPreview'); preview.src = URL.createObjectURL(file); preview.classList.remove('hidden'); // Enable transcribe button document.getElementById('transcribeBtn').disabled = false; } async function transcribeAudio() { if (!selectedFile) return; const loadingEl = document.getElementById('loadingIndicator'); const resultEl = document.getElementById('resultSection'); const transcribeBtn = document.getElementById('transcribeBtn'); // Show loading state loadingEl.classList.remove('hidden'); resultEl.classList.add('hidden'); transcribeBtn.disabled = true; try { const formData = new FormData(); formData.append('file', selectedFile); formData.append('format', 'json'); const startTime = Date.now(); const response = await fetch('/v1/audio/transcriptions', { method: 'POST', body: formData }); if (!response.ok) { throw new Error(`HTTP ${response.status}: ${response.statusText}`); } const result = await response.json(); const processingTime = ((Date.now() - startTime) / 1000).toFixed(1); // Display results document.getElementById('transcriptionText').value = result.text; document.getElementById('resultDuration').textContent = `Duration: ${result.duration.toFixed(1)}s`; document.getElementById('resultLanguage').textContent = `Language: ${result.language}`; resultEl.classList.remove('hidden'); console.log(`Transcription completed in ${processingTime}s`); } catch (error) { console.error('Transcription failed:', error); alert(`Transcription failed: ${error.message}`); } finally { loadingEl.classList.add('hidden'); transcribeBtn.disabled = false; } } function copyToClipboard() { const text = document.getElementById('transcriptionText').value; navigator.clipboard.writeText(text) .then(() => alert('Copied to clipboard')) .catch(err => console.error('Copy failed:', err)); } // Initialize window.addEventListener('load', () => { checkHealth(); loadSamplesList(); }); Key Takeaways and Production Considerations Building HIPAA-compliant voice-to-text systems requires architectural decisions that prioritize data privacy over convenience. The FLWhisper application demonstrates that you can achieve accurate medical transcription, fast processing times, and intuitive user experiences entirely on-premises. Critical lessons for healthcare AI: Privacy by architecture: Design systems where PHI never exists outside controlled environments, not as a configuration option No persistence by default: Audio and transcripts should be ephemeral unless explicitly saved with proper access controls Model selection matters: Whisper Medium provides medical terminology accuracy that smaller models miss Health checks enable reliability: Systems should verify model availability before accepting PHI Audit logging without content logging: Track operations for compliance without storing sensitive data in logs For production deployment in clinical settings, integrate with EHR systems via HL7/FHIR interfaces. Implement role-based access control with Active Directory integration. Add digital signatures for transcript authentication. Configure automatic PHI redaction using clinical NLP models. Deploy on HIPAA-compliant infrastructure with proper physical security. Implement comprehensive audit logging meeting compliance requirements. The complete implementation with ASP.NET Core API, Foundry Local integration, sample audio files, and comprehensive tests is available at github.com/leestott/FLWhisper. Clone the repository and follow the setup guide to experience privacy-first medical transcription. Resources and Further Reading FLWhisper Repository - Complete C# implementation with .NET 10 Quick Start Guide - Installation and usage instructions Microsoft Foundry Local Documentation - SDK reference and model catalog OpenAI Whisper Documentation - Model architecture and capabilities HIPAA Compliance Guidelines - HHS official guidance Testing Guide - Comprehensive test suite documentationBuilding a Smart Building HVAC Digital Twin with AI Copilot Using Foundry Local
Introduction Building operations teams face a constant challenge: optimizing HVAC systems for energy efficiency while maintaining occupant comfort and air quality. Traditional building management systems display raw sensor data, temperatures, pressures, CO₂ levels—but translating this into actionable insights requires deep HVAC expertise. What if operators could simply ask "Why is the third floor so warm?" and get an intelligent answer grounded in real building state? This article demonstrates building a sample smart building digital twin with an AI-powered operations copilot, implemented using DigitalTwin, React, Three.js, and Microsoft Foundry Local. You'll learn how to architect physics-based simulators that model thermal dynamics, implement 3D visualizations of building systems, integrate natural language AI control, and design fault injection systems for testing and training. Whether you're building IoT platforms for commercial real estate, designing energy management systems, or implementing predictive maintenance for building automation, this sample provides proven patterns for intelligent facility operations. Why Digital Twins Matter for Building Operations Physical buildings generate enormous operational data but lack intelligent interpretation layers. A 50,000 square foot office building might have 500+ sensors streaming metrics every minute, zone temperatures, humidity levels, equipment runtimes, energy consumption. Traditional BMS (Building Management Systems) visualize this data as charts and gauges, but operators must manually correlate patterns, diagnose issues, and predict failures. Digital twins solve this through physics-based simulation coupled with AI interpretation. Instead of just displaying current temperature readings, a digital twin models thermal dynamics, heat transfer rates, HVAC response characteristics, occupancy impacts. When conditions deviate from expectations, the twin compares observed versus predicted states, identifying root causes. Layer AI on top, and operators get natural language explanations: "The conference room is 3 degrees too warm because the VAV damper is stuck at 40% open, reducing airflow by 60%." This application focuses on HVAC, the largest building energy consumer, typically 40-50% of total usage. Optimizing HVAC by just 10% through better controls can save thousands of dollars monthly while improving occupant satisfaction. The digital twin enables "what-if" scenarios before making changes: "What happens to energy consumption and comfort if we raise the cooling setpoint by 2 degrees during peak demand response events?" Architecture: Three-Tier Digital Twin System The application implements a clean three-tier architecture separating visualization, simulation, and state management: The frontend uses React with Three.js for 3D visualization. Users see an interactive 3D model of the three-floor building with color-coded zones indicating temperature and CO₂ levels. Click any equipment, AHUs, VAVs, chillers, to see detailed telemetry. The control panel enables adjusting setpoints, running simulation steps, and activating demand response scenarios. Real-time charts display KPIs: energy consumption, comfort compliance, air quality levels. The backend Node.js/Express server orchestrates simulation and state management. It maintains the digital twin state as JSON, the single source of truth for all equipment, zones, and telemetry. REST API endpoints handle control requests, simulation steps, and AI copilot queries. WebSocket connections push real-time updates to the frontend for live monitoring. The HVAC simulator implements physics-based models: 1R1C thermal models for zones, affinity laws for fan power, chiller COP calculations, CO₂ mass balance equations. Foundry Local provides AI copilot capabilities. The backend uses foundry-local-sdk to query locally running models. Natural language queries ("How's the lobby temperature?") get answered with building state context. The copilot can explain anomalies, suggest optimizations, and even execute commands when explicitly requested. Implementing Physics-Based HVAC Simulation Accurate simulation requires modeling actual HVAC physics. The simulator implements several established building energy models: // backend/src/simulator/thermal-model.js class ZoneThermalModel { // 1R1C (one resistance, one capacitance) thermal model static calculateTemperatureChange(zone, delta_t_seconds) { const C_thermal = zone.volume * 1.2 * 1000; // Heat capacity (J/K) const R_thermal = zone.r_value * zone.envelope_area; // Thermal resistance // Internal heat gains (occupancy, equipment, lighting) const Q_internal = zone.occupancy * 100 + // 100W per person zone.equipment_load + zone.lighting_load; // Cooling/heating from HVAC const airflow_kg_s = zone.vav.airflow_cfm * 0.0004719; // CFM to kg/s const c_p_air = 1006; // Specific heat of air (J/kg·K) const Q_hvac = airflow_kg_s * c_p_air * (zone.vav.supply_temp - zone.temperature); // Envelope losses const Q_envelope = (zone.outdoor_temp - zone.temperature) / R_thermal; // Net energy balance const Q_net = Q_internal + Q_hvac + Q_envelope; // Temperature change: Q = C * dT/dt const dT = (Q_net / C_thermal) * delta_t_seconds; return zone.temperature + dT; } } This model captures essential thermal dynamics while remaining computationally fast enough for real-time simulation. It accounts for internal heat generation from occupants and equipment, HVAC cooling/heating contributions, and heat loss through the building envelope. The CO₂ model uses mass balance equations: class AirQualityModel { static calculateCO2Change(zone, delta_t_seconds) { // CO₂ generation from occupants const G_co2 = zone.occupancy * 0.0052; // L/s per person at rest // Outdoor air ventilation rate const V_oa = zone.vav.outdoor_air_cfm * 0.000471947; // CFM to m³/s // CO₂ concentration difference (indoor - outdoor) const delta_CO2 = zone.co2_ppm - 400; // Outdoor ~400ppm // Mass balance: dC/dt = (G - V*ΔC) / Volume const dCO2_dt = (G_co2 - V_oa * delta_CO2) / zone.volume; return zone.co2_ppm + (dCO2_dt * delta_t_seconds); } } These models execute every simulation step, updating the entire building state: async function simulateStep(twin, timestep_minutes) { const delta_t = timestep_minutes * 60; // Convert to seconds // Update each zone for (const zone of twin.zones) { zone.temperature = ZoneThermalModel.calculateTemperatureChange(zone, delta_t); zone.co2_ppm = AirQualityModel.calculateCO2Change(zone, delta_t); } // Update equipment based on zone demands for (const vav of twin.vavs) { updateVAVOperation(vav, twin.zones); } for (const ahu of twin.ahus) { updateAHUOperation(ahu, twin.vavs); } updateChillerOperation(twin.chiller, twin.ahus); updateBoilerOperation(twin.boiler, twin.ahus); // Calculate system KPIs twin.kpis = calculateSystemKPIs(twin); // Detect alerts twin.alerts = detectAnomalies(twin); // Persist updated state await saveTwinState(twin); return twin; } 3D Visualization with React and Three.js The frontend renders an interactive 3D building view that updates in real-time as conditions change. Using React Three Fiber simplifies Three.js integration with React's component model: // frontend/src/components/BuildingView3D.jsx import { Canvas } from '@react-three/fiber'; import { OrbitControls } from '@react-three/drei'; export function BuildingView3D({ twinState }) { return ( {/* Render building floors */} {twinState.zones.map(zone => ( selectZone(zone.id)} /> ))} {/* Render equipment */} {twinState.ahus.map(ahu => ( ))} ); } function ZoneMesh({ zone, onClick }) { const color = getTemperatureColor(zone.temperature, zone.setpoint); return ( ); } function getTemperatureColor(current, setpoint) { const deviation = current - setpoint; if (Math.abs(deviation) < 1) return '#00ff00'; // Green: comfortable if (Math.abs(deviation) < 3) return '#ffff00'; // Yellow: acceptable return '#ff0000'; // Red: uncomfortable } This visualization immediately shows building state at a glance, operators see "hot spots" in red, comfortable zones in green, and can click any area for detailed metrics. Integrating AI Copilot for Natural Language Control The AI copilot transforms building data into conversational insights. Instead of navigating multiple screens, operators simply ask questions: // backend/src/routes/copilot.js import { FoundryLocalClient } from 'foundry-local-sdk'; const foundry = new FoundryLocalClient({ endpoint: process.env.FOUNDRY_LOCAL_ENDPOINT }); router.post('/api/copilot/chat', async (req, res) => { const { message } = req.body; // Load current building state const twin = await loadTwinState(); // Build context for AI const context = buildBuildingContext(twin); const completion = await foundry.chat.completions.create({ model: 'phi-4', messages: [ { role: 'system', content: `You are an HVAC operations assistant for a 3-floor office building. Current Building State: ${context} Answer questions about equipment status, comfort conditions, and energy usage. Provide specific, actionable information based on the current data. Do not speculate beyond provided information.` }, { role: 'user', content: message } ], temperature: 0.3, max_tokens: 300 }); res.json({ response: completion.choices[0].message.content, model: 'phi-4', timestamp: new Date().toISOString() }); }); function buildBuildingContext(twin) { const alerts = twin.alerts.filter(a => a.active); const zoneStatus = twin.zones.map(z => `${z.name}: ${z.temperature.toFixed(1)}°F (setpoint: ${z.setpoint}°F), CO₂: ${z.co2_ppm}ppm, ${z.occupancy} people` ); return ` Zones: ${zoneStatus.join('\n')} Active Alerts: ${alerts.length} ${alerts.map(a => `- ${a.severity}: ${a.message}`).join('\n')} Energy: ${twin.kpis.total_power_kw.toFixed(1)} kW Chiller: ${twin.chiller.percent_load}% load, ${twin.chiller.cop.toFixed(2)} COP `.trim(); } Example interactions: User: "Why is the third floor conference room warm?" AI: "The Executive Suite on Floor 3 is currently 75.2°F, 3 degrees above the 72°F setpoint. This is due to high occupancy (8 people) and VAV-F3-EXEC operating at reduced airflow (285 CFM vs 450 CFM capacity). AHU-2 is supplying adequate cool air at 55°F. Recommend checking VAV damper position—it may be partially stuck." User: "What's our current energy usage?" AI: "Total building power consumption is 47.3 kW. Breakdown: Chiller (24.1 kW, 51%), AHU fans (12.8 kW, 27%), Pumps (7.2 kW, 15%), Other (3.2 kW, 7%). Chiller is operating at 62% load with a COP of 4.8, which is efficient for current conditions." User: "Activate demand response level 2" AI: "I cannot directly control building systems for safety reasons. To activate DR Level 2, use the Control Panel and click 'Demand Response' → 'Level 2'. This will raise cooling setpoints by 3°F and reduce auxiliary loads, targeting 15% energy reduction." The AI provides grounded, specific answers citing actual equipment IDs and metrics. It refuses to directly execute control commands, instead guiding operators to explicit control interfaces, a critical safety pattern for building systems. Fault Injection for Testing and Training Real building operations experience equipment failures, stuck dampers, sensor drift, communication losses. The digital twin includes comprehensive fault injection capabilities to train operators and test control logic: // backend/src/simulator/fault-injector.js const FAULT_CATALOG = { chillerFailure: { description: 'Chiller compressor failure', apply: (twin) => { twin.chiller.status = 'FAULT'; twin.chiller.cooling_output = 0; twin.alerts.push({ id: 'chiller-fault', severity: 'CRITICAL', message: 'Chiller compressor failure - no cooling available', equipment: 'CHILLER-01' }); } }, stuckVAVDamper: { description: 'VAV damper stuck at current position', apply: (twin, vavId) => { const vav = twin.vavs.find(v => v.id === vavId); vav.damper_stuck = true; vav.damper_position_fixed = vav.damper_position; twin.alerts.push({ id: `vav-stuck-${vavId}`, severity: 'HIGH', message: `VAV ${vavId} damper stuck at ${vav.damper_position}%`, equipment: vavId }); } }, sensorDrift: { description: 'Temperature sensor reading 5°F high', apply: (twin, zoneId) => { const zone = twin.zones.find(z => z.id === zoneId); zone.sensor_drift = 5.0; zone.temperature_measured = zone.temperature_actual + 5.0; } }, communicationLoss: { description: 'Equipment communication timeout', apply: (twin, equipmentId) => { const equipment = findEquipmentById(twin, equipmentId); equipment.comm_status = 'OFFLINE'; equipment.stale_data = true; twin.alerts.push({ id: `comm-loss-${equipmentId}`, severity: 'MEDIUM', message: `Lost communication with ${equipmentId}`, equipment: equipmentId }); } } }; router.post('/api/twin/fault', async (req, res) => { const { faultType, targetEquipment } = req.body; const twin = await loadTwinState(); const fault = FAULT_CATALOG[faultType]; if (!fault) { return res.status(400).json({ error: 'Unknown fault type' }); } fault.apply(twin, targetEquipment); await saveTwinState(twin); res.json({ message: `Applied fault: ${fault.description}`, affectedEquipment: targetEquipment, timestamp: new Date().toISOString() }); }); Operators can inject faults to practice diagnosis and response. Training scenarios might include: "The chiller just failed during a heat wave, how do you maintain comfort?" or "Multiple VAV dampers are stuck, which zones need immediate attention?" Key Takeaways and Production Deployment Building a physics-based digital twin with AI capabilities requires balancing simulation accuracy with computational performance, providing intuitive visualization while maintaining technical depth, and enabling AI assistance without compromising safety. Key architectural lessons: Physics models enable prediction: Comparing predicted vs observed behavior identifies anomalies that simple thresholds miss 3D visualization improves spatial understanding: Operators immediately see which floors or zones need attention AI copilots accelerate diagnosis: Natural language queries get answers in seconds vs. minutes of manual data examination Fault injection validates readiness: Testing failure scenarios prepares operators for real incidents JSON state enables integration: Simple file-based state makes connecting to real BMS systems straightforward For production deployment, connect the twin to actual building systems via BACnet, Modbus, or MQTT integrations. Replace simulated telemetry with real sensor streams. Calibrate model parameters against historical building performance. Implement continuous learning where the twin's predictions improve as it observes actual building behavior. The complete implementation with simulation engine, 3D visualization, AI copilot, and fault injection system is available at github.com/leestott/DigitalTwin. Clone the repository and run the startup scripts to explore the digital twin, no building hardware required. Resources and Further Reading Smart Building HVAC Digital Twin Repository - Complete source code and simulation engine Setup and Quick Start Guide - Installation instructions and usage examples Microsoft Foundry Local Documentation - AI integration reference HVAC Simulation Documentation - Physics model details and calibration Three.js Documentation - 3D visualization framework ASHRAE Standards - Building energy modeling standardsAgents League: Join the Reasoning Agents Track
In a previous blog post, we introduced Agents League, a two‑week AI agent challenge running February 16–27, and gave an overview of the three available tracks. In this post, we’ll zoom in on one of them in particular:🧠 The Reasoning Agents track, built on Microsoft Foundry. If you’re interested in multi‑step reasoning, planning, verification, and multi‑agent collaboration, this is the track designed for you. What Do We Mean by “Reasoning Agents”? Reasoning agents go beyond simple prompt–response interactions. They are agents that can: Plan how to approach a task Break problems into steps Reason across intermediate results Verify or critique their own outputs Collaborate with other agents to solve more complex problems With Microsoft Foundry (via UI or SDK) and/or the Microsoft Agent Framework, you can design agent systems that reflect real‑world decision‑making patterns—closer to how teams of humans work together. Why Build Reasoning Agents on Microsoft Foundry? Microsoft Foundry provides production‑ready building blocks for agentic systems, without locking you into a single way of working. For the Reasoning Agents track, Foundry enables you to: Define agent roles (planner, executor, verifier, critic, etc.) Orchestrate multi‑agent workflows Integrate tools, APIs, and MCP servers Apply structured reasoning patterns Observe and debug agent behavior as it runs You can work visually in the Foundry UI, programmatically via the SDK, or mix both approaches depending on your project. How to get started? Your first step to enter the arena is registering to the Agents League challenge: https://aka.ms/agentsleague/register. After you registered, navigate to the Reasoning Agent Starter Kit, to get more context about the challenge scenario, an example of multi-agent architecture to address it, along with some guidelines on the tech stack to use and useful resources to get started. There’s no single “correct” project, feel free to unleash your creativity and leverage AI-assisted development tools to accelerate your build process (e.g. GitHub Copilot). 👉 View the Reasoning Agents starter kit: https://github.com/microsoft/agentsleague/starter-kits Live Coding Battle: Reasoning Agents 📽️ Wednesday, Feb 18 – 9:00 AM PT During Week 1, we’re hosting a live coding battle dedicated entirely to the Reasoning Agents track. You’ll watch experienced developers from the community: Design agent architectures live Explain reasoning strategies and trade‑offs Make real‑time decisions about agent roles, tools, and flows The session is streamed on Microsoft Reactor and recorded, so you can watch it live (highly recommended for the best experience!) or later at your convenience. AMA Session on Discord 💬 Wednesday, Feb 25 – 9:00 AM PT In Week 2, it’s your turn to build—and ask questions. Join the Reasoning Agents AMA on Discord to: Ask about agent architecture and reasoning patterns Get clarification on Foundry capabilities Discuss MCP integration and multi‑agent design Get unstuck when your agent doesn’t behave as expected Prizes, Badges, and Recognition 🏆 $500 for the Reasoning Agents track winner 🎖️ Digital badge for everyone who registers and submits a project Important reminder: 👉 You must register before submitting to be eligible for prizes and the badge. Beyond the rewards, every participant receives feedback from Microsoft product teams, which is often the most valuable prize of all. Ready to Build Agents That Reason? If you’ve been curious about: Agentic architectures Multi‑step reasoning Verification and self‑reflection Building AI systems that explain their thinking …then the Reasoning Agents track is your arena. 📝 Register here: https://aka.ms/agentsleague/register 💬 Join Discord: https://aka.ms/agentsleague/discord 📽️ Watch live battles: https://aka.ms/agentsleague/battles The league starts February 16. The reasoning begins now.Agents League: Two Weeks, Three Tracks, One Challenge
We're inviting all developers to join Agents League, running February 16-27. It's a two-week challenge where you'll build AI agents using production-ready tools, learn from live coding sessions, and get feedback directly from Microsoft product teams. We've put together starter kits for each track to help you get up and running quickly that also includes requirements and guidelines. Whether you want to explore what GitHub Copilot can do beyond autocomplete, build reasoning agents on Microsoft Foundry, or create enterprise integrations for Microsoft 365 Copilot, we have a track for you. Important: Register first to be eligible for prizes and your digital badge. Without registration, you won't qualify for awards or receive a badge when you submit. What Is Agents League? It's a 2-week competition that combines learning with building: 📽️ Live coding battles – Watch Product teams, MVPs and community members tackle challenges in real-time on Microsoft Reactor 💻 Async challenges – Build at your own pace, on your schedule 💬 Discord community – Connect with other participants, join AMAs, and get help when you need it 🏆 Prizes – $500 per track winner, plus GitHub Copilot Pro subscriptions for top picks The Three Tracks 🎨 Creative Apps — Build with GitHub Copilot (Chat, CLI, or SDK) 🧠 Reasoning Agents — Build with Microsoft Foundry 💼 Enterprise Agents — Build with M365 Agents Toolkit (or Copilot Studio) More details on each track below, or jump straight to the starter kits. The Schedule Agents League starts on February 16th and runs through Feburary 27th. Within 2 weeks, we host live battles on Reactor and AMA sessions on Discord. Week 1: Live Battles (Feb 17-19) We're kicking off with live coding battles streamed on Microsoft Reactor. Watch experienced developers compete in real-time, explaining their approach and architectural decisions as they go. Tue Feb 17, 9 AM PT — 🎨 Creative Apps battle Wed Feb 18, 9 AM PT — 🧠 Reasoning Agents battle Thu Feb 19, 9 AM PT — 💼 Enterprise Agents battle All sessions are recorded, so you can watch on your own schedule. Week 2: Build + AMAs (Feb 24-26) This is your time to build and ask questions on Discord. The async format means you work when it suits you, evenings, weekends, whatever fits your schedule. We're also hosting AMAs on Discord where you can ask questions directly to Microsoft experts and product teams: Tue Feb 24, 9 AM PT — 🎨 Creative Apps AMA Wed Feb 25, 9 AM PT — 🧠 Reasoning Agents AMA Thu Feb 26, 9 AM PT — 💼 Enterprise Agents AMA Bring your questions, get help when you're stuck, and share what you're building with the community. Pick Your Track We've created a starter kit for each track with setup guides, project ideas, and example scenarios to help you get started quickly. 🎨 Creative Apps Tool: GitHub Copilot (Chat, CLI, or SDK) Build innovative, imaginative applications that showcase the potential of AI-assisted development. All application types are welcome, web apps, CLI tools, games, mobile apps, desktop applications, and more. The starter kit walks you through GitHub Copilot's different modes and provides prompting tips to get the best results. View the Creative Apps starter kit. 🧠 Reasoning Agents Tool: Microsoft Foundry (UI or SDK) and/or Microsoft Agent Framework Build a multi-agent system that leverages advanced reasoning capabilities to solve complex problems. This track focuses on agents that can plan, reason through multi-step problems, and collaborate. The starter kit includes architecture patterns, reasoning strategies (planner-executor, critic/verifier, self-reflection), and integration guides for tools and MCP servers. View the Reasoning Agents starter kit. 💼 Enterprise Agents Tool: M365 Agents Toolkit or Copilot Studio Create intelligent agents that extend Microsoft 365 Copilot to address real-world enterprise scenarios. Your agent must work on Microsoft 365 Copilot Chat. Bonus points for: MCP server integration, OAuth security, Adaptive Cards UI, connected agents (multi-agent architecture). View the Enterprise Agents starter kit. Prizes & Recognition To be eligible for prizes and your digital badge, you must register before submitting your project. Category Winners ($500 each): 🎨 Creative Apps winner 🧠 Reasoning Agents winner 💼 Enterprise Agents winner GitHub Copilot Pro subscriptions: Community Favorite (voted by participants on Discord) Product Team Picks (selected by Microsoft product teams) Everyone who registers and submits a project wins: A digital badge to showcase their participation. Beyond the prizes, every participant gets feedback from the teams who built these tools, a valuable opportunity to learn and improve your approach to AI agent development. How to Get Started Register first — This is required to be eligible for prizes and to receive your digital badge. Without registration, your submission won't qualify for awards or a badge. Pick a track — Choose one track. Explore the starter kits to help you decide. Watch the battles — See how experienced developers approach these challenges. Great for learning even if you're still deciding whether to compete. Build your project — You have until Feb 27. Work on your own schedule. Submit via GitHub — Open an issue using the project submission template. Join us on Discord — Get help, share your progress, and vote for your favorite projects on Discord. Links Register: https://aka.ms/agentsleague/register Starter Kits: https://github.com/microsoft/agentsleague/starter-kits Discord: https://aka.ms/agentsleague/discord Live Battles: https://aka.ms/agentsleague/battles Submit Project: Project submission templateAlphaLife Sciences powers regulatory-compliant AI workflows with PostgreSQL on Azure
by: Maxim Lukiyanov, PhD, Principal PM Manager and Sharon Chen, CEO and Founder at AlphaLife Sciences In life sciences, every document is deeply interconnected and highly regulated. Each clinical trial, regulatory submission, safety report, or protocol amendment is expected to stand up to rigorous audit. For AlphaLife Sciences, that challenge became an opportunity to rethink how AI could support expert human judgment. At Microsoft Ignite, AlphaLife Sciences CEO and Founder Sharon Chen shared how her team is building an AI-powered content authoring platform on top of Azure Database for PostgreSQL, designed specifically for the demands of regulated life sciences workflows. She also explained why the team is excited about Azure HorizonDB as a new PostgreSQL service that is built to meet the needs of modern enterprise workloads. This post explores how AlphaLife Sciences uses PostgreSQL as more than a data store. It’s a semantic foundation for compliant, auditable AI agents. Bringing AI into regulated workflows Life sciences organizations are under constant pressure. R&D pipelines are growing and patent windows are shrinking. A single clinical study report can take six months or more to complete, involving multiple teams and hundreds of source documents. Building efficiency into these processes is critical, but only if it doesn’t compromise accuracy, traceability, or compliance. That’s where many AI solutions fall short. Generating text is one thing, but generating verifiable, version-controlled, regulation-aware content is another. AlphaLife Sciences needed agents that could: Work across massive volumes of structured and unstructured data (Word, PDF, Excel, PowerPoint) Maintain full traceability from generated content back to source documents Support audits, amendments, and regulatory review Minimize hallucinations in a zero-tolerance environment Integrate naturally into the tools writers already use Bringing data, search, and AI together in one system At the core of AlphaLife Sciences’ platform is Azure Database for PostgreSQL. The team chose it for flexibility, extensibility, and for how well it supports modern AI workloads. Instead of stitching together separate systems for SQL queries, vector search, text indexing, and metadata tracking, AlphaLife Sciences consolidated everything into PostgreSQL. One of its flagship use cases is clinical trial protocol authoring, a process that typically involves: Designing trial objectives and endpoints Pulling references from previous studies Writing and revising hundreds of pages of structured content Managing multiple rounds of amendments and regulatory feedback With AI agents backed by PostgreSQL, that workflow changes dramatically. When a writer generates a protocol section, the system can automatically retrieve relevant references from a centralized document pool, using semantic search rather than manual lookup. Writers select the sources they want, apply rules or prompts, and let AI draft the section - complete with citations tied back to the original documents. Reviewers can inspect the source, adjust the output, or insert it directly into the document. For protocol amendments, the platform allows teams to upload inputs (Word or Excel), analyze which sections are affected, and generate structured suggestions. Changes are clearly highlighted, compared against previous versions, and summarized in amendment tables. AI agents that respect the rules A recurring theme in Chen’s talk was restraint. “We don’t just need AI that can write,” she said. “We need intelligent agents that understand data structures, follow regulatory laws, and manage version control.” This is where PostgreSQL-backed AI agents shine. By grounding AI behavior in structured schemas, controlled access, and auditable records, automation works hand-in-hand with human experts. AI accelerates first drafts, consistency checks, discrepancy detection, and cross-document analysis, but final accountability stays firmly with professionals. In some cases, the time to complete processes has been reduced by more than 50%. Azure Database for PostgreSQL has become more than a database for AlphaLife Sciences. It’s a semantic knowledge base that supports: Structured and unstructured data Vector similarity search Metadata-driven traceability Compliance, security, and auditability AI agents operating safely inside enterprise constraints By grounding AI agents directly in the database, reasoning, retrieval, and generation all operate against the same governed source of truth. “AI agents are not here to replace human beings,” said Chen. “They extend structured, compliant, and auditable thinking.” What’s next for AlphaLife Sciences with PostgreSQL on Azure Looking ahead, Chen shared her excitement about Azure HorizonDB and the capabilities it brings to PostgreSQL on Azure. Features like in-database AI model management, semantic operators for classification and summarization, and faster vector search with DiskANN align closely with AlphaLife Sciences’ needs as their platform continues to scale. “We’re extremely happy to see the launch of Azure HorizonDB and the more powerful tools coming with it,” Chen said. “By putting everything together in PostgreSQL, we don’t have to rely on different systems for vector search, text indexing, or SQL queries. Everything happens in one streamlined system. The code becomes cleaner, efficiency improves, and the AI agents perform much more elegantly.” Learn more AlphaLife Sciences’ journey was featured during the Microsoft Ignite session “The Blueprint for Intelligent AI Agents Backed by PostgreSQL.” Watch the session to learn more and see a demo of how Azure Database for PostgreSQL transforms the protocol and protocol amendment process. When AI is anchored in a strong PostgreSQL foundation, innovation and compliance don’t have to compete - they can reinforce each other.Rethinking Documentation Translation: Treating Translations as Versioned Software Assets
Rethinking Documentation Translation: Treating Translations as Versioned Software Assets This article is written from the perspective of maintaining large, open-source documentation repositories in the Microsoft ecosystem. I am the maintainer of Co-op Translator, an open-source tool for automating multilingual documentation translation, used across multiple large documentation repositories, including Microsoft’s For Beginners series. In large documentation repositories, translation problems rarely fail loudly. They fail quietly, and they accumulate over time. Recently, we made a fundamental design decision in how Co-op Translator handles translations. Translations are treated as versioned software assets, not static outputs. This article explains why we reached that conclusion, and what this perspective enables for teams maintaining large, fast-moving documentation repositories. When translations quietly become a liability In most documentation projects, translations are treated as finished outputs. Once a file is translated, it is assumed to remain valid until someone explicitly notices a problem. But documentation rarely stands still. Text changes. Code examples evolve. Screenshots are replaced. Notebooks are updated to reflect new behavior. The problem is that these changes are often invisible in translated content. A translation may still read fluently, while the information it contains is already out of date. At that point, the issue is no longer about translation quality. It becomes a maintenance problem. Reframing the question Most translation workflows implicitly ask: Is this translation correct? In practice, maintainers struggle with a different question: Is this translation still synchronized with the current source? This distinction matters. A translation can be correct and still be out of sync. Once we acknowledged this, it became clear that treating translations as static content was no longer sufficient. The design decision: translations as versioned assets Starting with Co-op Translator 0.16.2, we made a deliberate design decision: Translations are treated as versioned software assets. This applies not only to Markdown files, but also to images, notebooks, and any other translated artifacts. Translated content is not just text. It is an artifact generated from a specific version of a source. To make this abstraction operational rather than theoretical, we did not invent a new mechanism. Instead, we looked to systems that already solve a similar problem: pip, poetry, and npm. These tools are designed to track artifacts as their sources evolve. We applied the same thinking to translated content. Closer to dependency management than translation jobs The closest analogy is software dependency management. When a dependency becomes outdated: it is not suddenly “wrong,” it is simply no longer aligned with the current version. Translations behave the same way. When the source document changes: the translated file does not immediately become incorrect, it becomes out of sync with its source version. This framing shifts the problem away from translation output and toward state and synchronization. Why file-level versioning matters Many translation systems operate at the string or segment level. That model works well for UI text and relatively stable resources. Documentation is different. A Markdown file is an artifact. A screenshot is an artifact. A notebook is an artifact. They are consumed as units, not as isolated strings. Managing translation state at the file level allows maintainers to reason about translations using the same mental model they already apply to other repository assets. What changed in practice From embedded markers to explicit state Previously, translation metadata lived inside translated files as embedded comments or markers. This approach had clear limitations: translation state was fragmented, difficult to inspect globally, and easy to miss as repositories grew. We moved to language-scoped JSON state files that explicitly track: the source version, the translated artifact, and its synchronization status. Translation state is no longer hidden inside content. It is a first-class, inspectable part of the repository. Extending the model to images and notebooks The same model now applies consistently to: translated images, localized notebooks, and other non-text artifacts. If an image changes in the source language, the translated image becomes out of sync. If a notebook is updated, its translated versions are evaluated against the new source version. The format does not matter. The lifecycle does. Once translations are treated as versioned assets, the system remains consistent across all content types. What this enables This design enables: Explicit drift detection See which translations are out of sync without guessing. Consistent maintenance signals Text, images, and notebooks follow the same rules. Clear responsibility boundaries The system reports state. Humans decide action. Scalability for fast-moving repositories Translation maintenance becomes observable, not reactive. In large documentation sets, this difference determines whether translation maintenance is sustainable at all. What this is not This system does not: judge translation quality, determine semantic correctness, or auto-approve content. It answers one question only: Is this translated artifact synchronized with its source version? Who this is for This approach is designed for teams that: maintain multilingual documentation, update content frequently, and need confidence in what is actually up to date. When documentation evolves faster than translations, treating translations as versioned assets becomes a necessity, not an optimization. Closing thought Once translations are modeled as software assets, long-standing ambiguities disappear. State becomes visible. Maintenance becomes manageable. And translations fit naturally into existing software workflows. At that point, the question is no longer whether translation drift exists, but: Can you see it? Reference Co-op Translator repository https://github.com/Azure/co-op-translatorHow to Build Safe Natural Language-Driven APIs
TL;DR Building production natural language APIs requires separating semantic parsing from execution. Use LLMs to translate user text into canonical structured requests (via schemas), then execute those requests deterministically. Key patterns: schema completion for clarification, confidence gates to prevent silent failures, code-based ontologies for normalization, and an orchestration layer. This keeps language as input, not as your API contract. Introduction APIs that accept natural language as input are quickly becoming the norm in the age of agentic AI apps and LLMs. From search and recommendations to workflows and automation, users increasingly expect to "just ask" and get results. But treating natural language as an API contract introduces serious risks in production systems: Nondeterministic behavior Prompt-driven business logic Difficult debugging and replay Silent failures that are hard to detect In this post, I'll describe a production-grade architecture for building safe, natural language-driven APIs: one that embraces LLMs for intent discovery and entity extraction while preserving the determinism, observability, and reliability that backend systems require. This approach is based on building real systems using Azure OpenAI and LangGraph, and on lessons learned the hard way. The Core Problem with Natural Language APIs Natural language is an excellent interface for humans. It is a poor interface for systems. When APIs accept raw text directly and execute logic based on it, several problems emerge: The API contract becomes implicit and unversioned Small prompt changes cause behavioral changes Business logic quietly migrates into prompts In short: language becomes the contract, and that's fragile. The solution is not to avoid natural language, but to contain it. A Key Principle: Natural Language Is Input, Not a Contract So how do we contain it? The answer lies in treating natural language fundamentally differently than we treat traditional API inputs. The most important design decision we made was this: Natural language should be translated into structure, not executed directly. That single principle drives the entire architecture. Instead of building "chatty APIs," we split responsibilities clearly: Natural language is used for intent discovery and entity extraction Structured data is used for execution Two Explicit API Layers This principle translates into a concrete architecture with two distinct API layers, each with a single, clear responsibility. 1. Semantic Parse API (Natural Language → Structure) This API: Accepts user text Extracts intent and entities using LLMs Completes a predefined schema Asks clarifying questions when required Returns a canonical, structured request Does not execute business logic Think of this as a compiler, not an engine. 2. Structured Execution API (Structure → Action) This API: Accepts only structured input Calls downstream systems to process the request and get results Is deterministic and versioned Contains no natural language handling Is fully testable and replayable This is where execution happens. Why This Separation Matters Separating these layers gives you: A stable, versionable API contract Freedom to improve NLP without breaking clients Clear ownership boundaries Deterministic execution paths Most importantly, it prevents LLM behavior from leaking into core business logic. Canonical Schemas Are the Backbone Now that we've established the two-layer architecture, let's dive into what makes it work: canonical schemas. Each supported intent is defined by a canonical schema that lives in code. Example (simplified): This schema is used when a user is looking for similar product recommendations. The entities capture which product to use as reference and how to bias the recommendations toward price or quality. { "intent": "recommend_similar", "entities": { "reference_product_id": "string", "price_bias": "number (-1 to 1)", "quality_bias": "number (-1 to 1)" } } Schemas define: Required vs optional fields Allowed ranges and types Validation rules They are the contract, not the prompt. When a user says "show me products like the blue backpack but cheaper", the LLM extracts: Intent: recommend_similar reference_product_id: "blue_backpack_123" price_bias: -0.8 (strongly prefer cheaper) quality_bias: 0.0 (neutral) The schema ensures that even if the user phrased it as "find alternatives to item 123 with better pricing" or "cheaper versions of that blue bag", the output is always the same structure. The natural language variation is absorbed at the semantic layer. The execution layer receives a consistent, validated request every time. This decoupling is what makes the system maintainable. Schema Completion, Not Free-Form Chat But what happens when the user's input doesn't contain all the information needed to complete the schema? This is where structured clarification comes in. A common misconception is that clarification means "chatting until it feels right." In production systems, clarification is schema completion. If required fields are missing or ambiguous, the semantic API responds with: What information is missing A targeted clarification question The current schema state Example response: { "status": "needs_clarification", "missing_fields": ["reference_product_id"], "question": "Which product should I compare against?", "state": { "intent": "recommend_similar", "entities": { "reference_product_id": null, "price_bias": -0.3, "quality_bias": 0.4 } } } The state object is the memory. The API itself remains stateless. A Complete Conversation Flow To illustrate how schema completion works in practice, here's a full conversation flow where the user's initial request is missing required information: Initial Request: User: "Show me cheaper alternatives with good quality" API Response (needs clarification): { "status": "needs_clarification", "missing_fields": ["reference_product_id"], "question": "Which product should I compare against?", "state": { "intent": "recommend_similar", "entities": { "reference_product_id": null, "price_bias": -0.3, "quality_bias": 0.4 } } } Follow-up Request: User: "The blue backpack" Client sends: { "user_input": "The blue backpack", "state": { "intent": "recommend_similar", "entities": { "reference_product_id": null, "price_bias": -0.3, "quality_bias": 0.4 } } } API Response (complete): { "status": "complete", "canonical_request": { "intent": "recommend_similar", "entities": { "reference_product_id": "blue_backpack_123", "price_bias": -0.3, "quality_bias": 0.4 } } } The client passes the state back with each clarification. The API remains stateless, while the client manages the conversation context. Once complete, the canonical_request can be sent directly to the execution API. Why LangGraph Fits This Problem Perfectly With schemas and clarification flows defined, we need a way to orchestrate the semantic parsing workflow reliably. This is where LangGraph becomes valuable. LangGraph allows semantic parsing to be modeled as a structured, deterministic workflow with explicit decision points: Classify intent: Determine what the user wants to do from a predefined set of supported actions Extract candidate entities: Pull out relevant parameters from the natural language input using the LLM Merge into schema state: Map the extracted values into the canonical schema structure Validate required fields: Check if all mandatory fields are present and values are within acceptable ranges Either complete or request clarification: Return the canonical request if complete, or ask a targeted question if information is missing Each node has a single responsibility. Validation and routing are done in code, not by the LLM. LangGraph provides: Explicit state transitions Deterministic routing Observable execution Safe retries Used this way, it becomes a powerful orchestration tool, not a conversational agent. Confidence Gates Prevent Silent Failures Structured workflows handle the process, but there's another critical safety mechanism we need: knowing when the LLM isn't confident about its extraction. Even when outputs are structurally valid, they may not be reliable. We require the semantic layer to emit a confidence score. If confidence falls below a threshold, execution is blocked and clarification is requested. This simple rule eliminates an entire class of silent misinterpretations that are otherwise very hard to detect. Example: When a user says "Show me items similar to the bag", the LLM might extract: { "intent": "recommend_similar", "confidence": 0.55, "entities": { "reference_product_id": "generic_bag_001", "confidence_scores": { "reference_product_id": 0.4 } } } The overall confidence is low (0.55), and the entity confidence for reference_product_id is very low (0.4) because "the bag" is ambiguous. There might be hundreds of bags in the catalog. Instead of proceeding with a potentially wrong guess, the API responds: { "status": "needs_clarification", "reason": "low_confidence", "question": "I found multiple bags. Did you mean the blue backpack, the leather tote, or the travel duffel?", "confidence": 0.55 } This prevents the system from silently executing the wrong recommendation and provides a better user experience. Lightweight Ontologies (Keep Them in Code) Beyond confidence scoring, we need a way to normalize the variety of terms users might use into consistent canonical values. We also introduced lightweight, code-level ontologies: Allowed intents Required entities per intent Synonym-to-canonical mappings Cross-field validation rules These live in code and configuration, not in prompts. LLMs propose values. Code enforces meaning. Example: Consider these user inputs that all mean the same thing: "Show me cheaper options" "Find budget-friendly alternatives" "I want something more affordable" "Give me lower-priced items" The LLM might extract different values: "cheaper", "budget-friendly", "affordable", "lower-priced". The ontology maps all of these to a canonical value: PRICE_BIAS_SYNONYMS = { "cheaper": -0.7, "budget-friendly": -0.7, "affordable": -0.7, "lower-priced": -0.7, "expensive": 0.7, "premium": 0.7, "high-end": 0.7 } When the LLM extracts "budget-friendly", the code normalizes it to -0.7 for the price_bias field. Similarly, cross-field validation catches logical inconsistencies: if entities["price_bias"] < -0.5 and entities["quality_bias"] > 0.5: return clarification("You want cheaper items with higher quality. This might be difficult. Should I prioritize price or quality?") The LLM proposes. The ontology normalizes. The validation enforces business rules. What About Latency? A common concern with multi-step semantic parsing is performance. In practice, we observed: Intent classification: ~40 ms Entity extraction: ~200 ms Validation and routing: ~1 ms Total overhead: ~250–300 ms. For chat-driven user experiences, this is well within acceptable bounds and far cheaper than incorrect or inconsistent execution. Key Takeaways Let's bring it all together. If you're building APIs that accept natural language in production: Do not make language your API contract Translate language into canonical structure Own schema completion server-side Use LLMs for discovery and extraction, not execution Treat safety and determinism as first-class requirements Natural language is an input format. Structure is the contract. Closing Thoughts LLMs make it easy to build impressive demos. Building safe, reliable systems with them requires discipline. By separating semantic interpretation from execution, and by using tools like Azure OpenAI and LangGraph thoughtfully, you can build natural language-driven APIs that scale, evolve, and behave predictably in production. Hopefully, this architecture saves you a few painful iterations.Benchmarking Local AI Models
Introduction Selecting the right AI model for your application requires more than reading benchmark leaderboards. Published benchmarks measure academic capabilities, question answering, reasoning, coding, but your application has specific requirements: latency budgets, hardware constraints, quality thresholds. How do you know if Phi-4 provides acceptable quality for your document summarization use case? Will Qwen2.5-0.5B meet your 100ms response time requirement? Does your edge device have sufficient memory for Phi-3.5 Mini? The answer lies in empirical testing: running actual models on your hardware with your workload patterns. This article demonstrates building a comprehensive model benchmarking platform using FLPerformance, Node.js, React, and Microsoft Foundry Local. You'll learn how to implement scientific performance measurement, design meaningful benchmark suites, visualize multi-dimensional comparisons, and make data-driven model selection decisions. Whether you're evaluating models for production deployment, optimizing inference costs, or validating hardware specifications, this platform provides the tools for rigorous performance analysis. Why Model Benchmarking Requires Purpose-Built Tools You cannot assess model performance by running a few manual tests and noting the results. Scientific benchmarking demands controlled conditions, statistically significant sample sizes, multi-dimensional metrics, and reproducible methodology. Understand why purpose-built tooling is essential. Performance is multi-dimensional. A model might excel at throughput (tokens per second) but suffer at latency (time to first token). Another might generate high-quality outputs slowly. Your application might prioritize consistency over average performance, a model with variable response times (high p95/p99 latency) creates poor user experiences even if averages look good. Measuring all dimensions simultaneously enables informed tradeoffs. Hardware matters enormously. Benchmark results from NVIDIA A100 GPUs don't predict performance on consumer laptops. NPU acceleration changes the picture again. Memory constraints affect which models can even load. Test on your actual deployment hardware or comparable specifications to get actionable results. Concurrency reveals bottlenecks. A model handling one request excellently might struggle with ten concurrent requests. Real applications experience variable load, measuring only single-threaded performance misses critical scalability constraints. Controlled concurrency testing reveals these limits. Statistical rigor prevents false conclusions. Running a prompt once and noting the response time tells you nothing about performance distribution. Was this result typical? An outlier? You need dozens or hundreds of trials to establish p50/p95/p99 percentiles, understand variance, and detect stability issues. Comparison requires controlled experiments. Different prompts, different times of day, different system loads, all introduce confounding variables. Scientific comparison runs identical workloads across models sequentially, controlling for external factors. Architecture: Three-Layer Performance Testing Platform FLPerformance implements a clean separation between orchestration, measurement, and presentation: The frontend React application provides model management, benchmark configuration, test execution, and results visualization. Users add models from the Foundry Local catalog, configure benchmark parameters (iterations, concurrency, timeout values), launch test runs, and view real-time progress. The results dashboard displays comparison tables, latency distribution charts, throughput graphs, and "best model for..." recommendations. The backend Node.js/Express server orchestrates tests and captures metrics. It manages the single Foundry Local service instance, loads/unloads models as needed, executes benchmark suites with controlled concurrency, measures comprehensive metrics (TTFT, TPOT, total latency, throughput, error rates), and persists results to JSON storage. WebSocket connections provide real-time progress updates during long benchmark runs. Foundry Local SDK integration uses the official foundry-local-sdk npm package. The SDK manages service lifecycle, starting, stopping, health checkin, and handles model operations, downloading, loading into memory, unloading. It provides OpenAI-compatible inference APIs for consistent request formatting across models. The architecture supports simultaneous testing of multiple models by loading them one at a time, running identical benchmarks, and aggregating results for comparison: User Initiates Benchmark Run ↓ Backend receives {models: [...], suite: "default", iterations: 10} ↓ For each model: 1. Load model into Foundry Local 2. Execute benchmark suite - For each prompt in suite: * Run N iterations * Measure TTFT, TPOT, total time * Track errors and timeouts * Calculate tokens/second 3. Aggregate statistics (mean, p50, p95, p99) 4. Unload model ↓ Store results with metadata ↓ Return comparison data to frontend ↓ Visualize performance metrics Implementing Scientific Measurement Infrastructure Accurate performance measurement requires instrumentation that captures multiple dimensions without introducing measurement overhead: // src/server/benchmark.js import { performance } from 'perf_hooks'; export class BenchmarkExecutor { constructor(foundryClient, options = {}) { this.client = foundryClient; this.options = { iterations: options.iterations || 10, concurrency: options.concurrency || 1, timeout_ms: options.timeout_ms || 30000, warmup_iterations: options.warmup_iterations || 2 }; } async runBenchmarkSuite(modelId, prompts) { const results = []; // Warmup phase (exclude from results) console.log(`Running ${this.options.warmup_iterations} warmup iterations...`); for (let i = 0; i < this.options.warmup_iterations; i++) { await this.executePrompt(modelId, prompts[0].text); } // Actual benchmark runs for (const prompt of prompts) { console.log(`Benchmarking prompt: ${prompt.id}`); const measurements = []; for (let i = 0; i < this.options.iterations; i++) { const measurement = await this.executeMeasuredPrompt( modelId, prompt.text ); measurements.push(measurement); // Small delay between iterations to stabilize await sleep(100); } results.push({ prompt_id: prompt.id, prompt_text: prompt.text, measurements, statistics: this.calculateStatistics(measurements) }); } return { model_id: modelId, timestamp: new Date().toISOString(), config: this.options, results }; } async executeMeasuredPrompt(modelId, promptText) { const measurement = { success: false, error: null, ttft_ms: null, // Time to first token tpot_ms: null, // Time per output token total_ms: null, tokens_generated: 0, tokens_per_second: 0 }; try { const startTime = performance.now(); let firstTokenTime = null; let tokenCount = 0; // Streaming completion to measure TTFT const stream = await this.client.chat.completions.create({ model: modelId, messages: [{ role: 'user', content: promptText }], max_tokens: 200, temperature: 0.7, stream: true }); for await (const chunk of stream) { if (chunk.choices[0]?.delta?.content) { if (firstTokenTime === null) { firstTokenTime = performance.now(); measurement.ttft_ms = firstTokenTime - startTime; } tokenCount++; } } const endTime = performance.now(); measurement.total_ms = endTime - startTime; measurement.tokens_generated = tokenCount; if (tokenCount > 1 && firstTokenTime) { // TPOT = time after first token / (tokens - 1) const timeAfterFirstToken = endTime - firstTokenTime; measurement.tpot_ms = timeAfterFirstToken / (tokenCount - 1); measurement.tokens_per_second = 1000 / measurement.tpot_ms; } measurement.success = true; } catch (error) { measurement.error = error.message; measurement.success = false; } return measurement; } calculateStatistics(measurements) { const successful = measurements.filter(m => m.success); const total = measurements.length; if (successful.length === 0) { return { success_rate: 0, error_rate: 1.0, sample_size: total }; } const ttfts = successful.map(m => m.ttft_ms).sort((a, b) => a - b); const tpots = successful.map(m => m.tpot_ms).filter(v => v !== null).sort((a, b) => a - b); const totals = successful.map(m => m.total_ms).sort((a, b) => a - b); const throughputs = successful.map(m => m.tokens_per_second).filter(v => v > 0); return { success_rate: successful.length / total, error_rate: (total - successful.length) / total, sample_size: total, ttft: { mean: mean(ttfts), median: percentile(ttfts, 50), p95: percentile(ttfts, 95), p99: percentile(ttfts, 99), min: Math.min(...ttfts), max: Math.max(...ttfts) }, tpot: tpots.length > 0 ? { mean: mean(tpots), median: percentile(tpots, 50), p95: percentile(tpots, 95) } : null, total_latency: { mean: mean(totals), median: percentile(totals, 50), p95: percentile(totals, 95), p99: percentile(totals, 99) }, throughput: { mean_tps: mean(throughputs), median_tps: percentile(throughputs, 50) } }; } } function mean(arr) { return arr.reduce((sum, val) => sum + val, 0) / arr.length; } function percentile(sortedArr, p) { const index = Math.ceil((sortedArr.length * p) / 100) - 1; return sortedArr[Math.max(0, index)]; } function sleep(ms) { return new Promise(resolve => setTimeout(resolve, ms)); } This measurement infrastructure captures: Time to First Token (TTFT): Critical for perceived responsiveness—users notice delays before output begins Time Per Output Token (TPOT): Determines generation speed after first token—affects throughput Total latency: End-to-end time—matters for batch processing and high-volume scenarios Tokens per second: Overall throughput metric—useful for capacity planning Statistical distributions: Mean alone masks variability—p95/p99 reveal tail latencies that impact user experience Success/error rates: Stability metrics—some models timeout or crash under load Designing Meaningful Benchmark Suites Benchmark quality depends on prompt selection. Generic prompts don't reflect real application behavior. Design suites that mirror actual use cases: // benchmarks/suites/default.json { "name": "default", "description": "General-purpose benchmark covering diverse scenarios", "prompts": [ { "id": "short-factual", "text": "What is the capital of France?", "category": "factual", "expected_tokens": 5 }, { "id": "medium-explanation", "text": "Explain how photosynthesis works in 3-4 sentences.", "category": "explanation", "expected_tokens": 80 }, { "id": "long-reasoning", "text": "Analyze the economic factors that led to the 2008 financial crisis. Discuss at least 5 major causes with supporting details.", "category": "reasoning", "expected_tokens": 250 }, { "id": "code-generation", "text": "Write a Python function that finds the longest palindrome in a string. Include docstring and example usage.", "category": "coding", "expected_tokens": 150 }, { "id": "creative-writing", "text": "Write a short story (3 paragraphs) about a robot learning to paint.", "category": "creative", "expected_tokens": 200 } ] } This suite covers multiple dimensions: Length variation: Short (5 tokens), medium (80), long (250)—tests models across output ranges Task diversity: Factual recall, explanation, reasoning, code, creative—reveals capability breadth Token predictability: Expected token counts enable throughput calculations For production applications, create custom suites matching your actual workload: { "name": "customer-support", "description": "Simulates actual customer support queries", "prompts": [ { "id": "product-question", "text": "How do I reset my password for the customer portal?" }, { "id": "troubleshooting", "text": "I'm getting error code 503 when trying to upload files. What should I do?" }, { "id": "policy-inquiry", "text": "What is your refund policy for annual subscriptions?" } ] } Visualizing Multi-Dimensional Performance Comparisons Raw numbers don't reveal insights—visualization makes patterns obvious. The frontend implements several comparison views: Comparison Table shows side-by-side metrics: // frontend/src/components/ResultsTable.jsx export function ResultsTable({ results }) { return ( {results.map(result => ( ))} Model TTFT (ms) TPOT (ms) Throughput (tok/s) P95 Latency Error Rate {result.model_id} {result.stats.ttft.median.toFixed(0)} (p95: {result.stats.ttft.p95.toFixed(0)}) {result.stats.tpot?.median.toFixed(1) || 'N/A'} {result.stats.throughput.median_tps.toFixed(1)} {result.stats.total_latency.p95.toFixed(0)} ms 0.05 ? 'error' : 'success'}> {(result.stats.error_rate * 100).toFixed(1)}% ); } Latency Distribution Chart reveals performance consistency: // Using Chart.js for visualization export function LatencyChart({ results }) { const data = { labels: results.map(r => r.model_id), datasets: [ { label: 'Median (p50)', data: results.map(r => r.stats.total_latency.median), backgroundColor: 'rgba(75, 192, 192, 0.5)' }, { label: 'p95', data: results.map(r => r.stats.total_latency.p95), backgroundColor: 'rgba(255, 206, 86, 0.5)' }, { label: 'p99', data: results.map(r => r.stats.total_latency.p99), backgroundColor: 'rgba(255, 99, 132, 0.5)' } ] }; return ( ); } Recommendations Engine synthesizes multi-dimensional comparison: export function generateRecommendations(results) { const recommendations = []; // Find fastest TTFT (best perceived responsiveness) const fastestTTFT = results.reduce((best, r) => r.stats.ttft.median < best.stats.ttft.median ? r : best ); recommendations.push({ category: 'Fastest Response', model: fastestTTFT.model_id, reason: `Lowest median TTFT: ${fastestTTFT.stats.ttft.median.toFixed(0)}ms` }); // Find highest throughput const highestThroughput = results.reduce((best, r) => r.stats.throughput.median_tps > best.stats.throughput.median_tps ? r : best ); recommendations.push({ category: 'Best Throughput', model: highestThroughput.model_id, reason: `Highest tok/s: ${highestThroughput.stats.throughput.median_tps.toFixed(1)}` }); // Find most consistent (lowest p95-p50 spread) const mostConsistent = results.reduce((best, r) => { const spread = r.stats.total_latency.p95 - r.stats.total_latency.median; const bestSpread = best.stats.total_latency.p95 - best.stats.total_latency.median; return spread < bestSpread ? r : best; }); recommendations.push({ category: 'Most Consistent', model: mostConsistent.model_id, reason: 'Lowest latency variance (p95-p50 spread)' }); return recommendations; } Key Takeaways and Benchmarking Best Practices Effective model benchmarking requires scientific methodology, comprehensive metrics, and application-specific testing. FLPerformance demonstrates that rigorous performance measurement is accessible to any development team. Critical principles for model evaluation: Test on target hardware: Results from cloud GPUs don't predict laptop performance Measure multiple dimensions: TTFT, TPOT, throughput, consistency all matter Use statistical rigor: Single runs mislead—capture distributions with adequate sample sizes Design realistic workloads: Generic benchmarks don't predict your application's behavior Include warmup iterations: Model loading and JIT compilation affect early measurements Control concurrency: Real applications handle multiple requests—test at realistic loads Document methodology: Reproducible results require documented procedures and configurations The complete benchmarking platform with model management, measurement infrastructure, visualization dashboards, and comprehensive documentation is available at github.com/leestott/FLPerformance. Clone the repository and run the startup script to begin evaluating models on your hardware. Resources and Further Reading FLPerformance Repository - Complete benchmarking platform Quick Start Guide - Setup and first benchmark run Microsoft Foundry Local Documentation - SDK reference and model catalog Architecture Guide - System design and SDK integration Benchmarking Best Practices - Methodology and troubleshooting
