Recent Discussions
Azure AI foundry SDK-Tool Approval Not Triggered When Using ConnectedAgentTool() Between Agents
I am building an orchestration workflow in Azure AI Foundry using the Python SDK. Each agent uses tools exposed via an MCP server (deployed in Azure container app), and individual agents work perfectly when run independently — tool approval is triggered, and execution proceeds as expected. I have a main agent which orchestrates the flow of these individual agents.However, when I connect one agent to another using ConnectedAgentTool(), the tool approval flow never occurs, and orchestration halts. All I see is the run status as IN-PROGRESS for some time and then exits. The downstream (child) agent is never invoked. I have tried mcp_tool.set_approval_mode("never") , but that didn't help. Auto-Approval Implementation: I have implemented a polling loop that checks the run status and auto-approves any requires_action events. async def poll_run_until_complete(project_client: AIProjectClient, thread_id: str, run_id: str): """ Polls the run until completion. Auto-approves any tool calls encountered. """ while True: run = await project_client.agents.runs.get(thread_id=thread_id, run_id=run_id) status = getattr(run, "status", None) print(f"[poll] Run {run_id} status: {status}") # Completed states if status in ("succeeded", "failed", "cancelled", "completed"): print(f"[poll] Final run status: {status}") if status == "failed": print("Run last_error:", getattr(run, "last_error", None)) return run # Auto-approve any tool calls if status == "requires_action" and isinstance(getattr(run, "required_action", None), SubmitToolApprovalAction): submit_action = run.required_action.submit_tool_approval tool_calls = getattr(submit_action, "tool_calls", []) or [] if not tool_calls: print("[poll] requires_action but no tool_calls found. Waiting...") else: approvals = [] for tc in tool_calls: print(f"[poll] Auto-approving tool call: {tc.id} name={tc.name} args={tc.arguments}") approvals.append(ToolApproval(tool_call_id=tc.id, approve=True)) if approvals: await project_client.agents.runs.submit_tool_outputs( thread_id=thread_id, run_id=run_id, tool_approvals=approvals ) print("[poll] Submitted tool approvals.") else: # Debug: Inspect run steps if stuck run_steps = [s async for s in project_client.agents.run_steps.list(thread_id=thread_id, run_id=run_id)] if run_steps: for step in run_steps: sid = getattr(step, "id", None) sstatus = getattr(step, "status", None) print(f" step: id={sid} status={sstatus}") step_details = getattr(step, "step_details", None) if step_details: tool_calls = getattr(step_details, "tool_calls", None) if tool_calls: for call in tool_calls: print(f" tool_call id={getattr(call,'id',None)} name={getattr(call,'name',None)} args={getattr(call,'arguments',None)} output={getattr(call,'output',None)}") await asyncio.sleep(1) This code works and auto-approves tool calls for MCP tools. But while using ConnectedAgentTool(), the run never enters requires_action — so no approvals are requested, and the orchestration halts. Environment: azure-ai-agents==1.2.0b4 azure-ai-projects==1.1.0b4 Python: 3.11.13 Auth: DefaultAzureCredential Notes: MCP tools work and trigger approval normally when directly attached. and I ndividual agents function as expected in standalone runs. Can anyone help here..!
Issue when connecting from SPFX to Entra-enabled Azure AI Foundry resource
We have been successfully connecting our chat bot from an SPFX to a chat completion model in Azure, using key authentication. We have a requirement now to disable key authentication. This is what we've done so far: disabled API authentication in the resource Gave to the SharePoint Client Extensibility Web Application Principal "Cognitive Services OpenAI User", "Cognitive Service User" and "Cognitive Data Reader" permission in the resource In the SPFX we have added the following in the package-solution.json (and we have approved it in the SharePoint admin site): "webApiPermissionRequests": [ { "resource": "Azure Machine Learning Services", "scope": "user_impersonation" } ] To connect to the chat completion API we're using fetchEventSource from '@microsoft/fetch-event-source', so we're getting a Bearer token using AadTokenProviderFactory from "@microsoft/sp-http", e.g.: // preceeded by some code to get the tokenProvider from aadTokenProviderFactory const token = await tokenProvider.getToken('https://ai.azure.com'); const url = "https://my-ai-resource.openai.azure.com/openai/deployments/gpt-4o/chat/completions?api-version=2025-01-01-preview"; await fetchEventSource(url, { method: 'POST', headers: { Accept: 'text/event-stream', 'Content-type': 'application/json', Authorization: `Bearer ${token}` }, body: body, ...// truncated We added the users (let's say, email address removed for privacy reasons) in the resource as an Azure AI User. When we try to get this to work, we get the following error: The principal `email address removed for privacy reasons` lacks the required data action `Microsoft.CognitiveServices/accounts/OpenAI/deployments/chat/completions/action` to perform `POST /openai/deployments/{deployment-id}/chat/completions` operation. How can we make this work? Ideally we would prefer the SPFX principal to do the request to the chat completion API, without needed to have to add end users in the resource thorugh IAC, but my understanding is that AadTokenProviderFactory only issues delegated access tokens.
Responses API for gpt-4.1 in Europe
Hello everyone! I'm writing here trying to figure out something about the availability of the "responses" APIs in european regions: https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/responses?tabs=python-key i'm trying to deploy a /responses endpoint for the model we are currently using, gpt-4.1, since i've read that the /completions endpoint will be dismissed by OpenAI starting from august 2026. Our application is currently migrating all the API calls from completions to responses, and we were wondering if we could already do the same for our clients in Europe as well, which have to comply to GDPR and currently use our Azure subscription. In the page linked above, i can see some regions that would fit our needs, in particular: francecentral norwayeast polandcentral swedencentral switzerlandnorth but then, i can also read "Not every model is available in the regions supported by the responses API.", which probably already answers my question: from the Azure AI Foundry Portal, i wasn't able to deploy such endpoint in those regions, except for the o3 model. For the 4.1 model, only the completions endpoint is listed, while searching for "Responses" (in "Deploy base model") returns only o3 and these others: Can you confirm that i'm not doing anything wrong (looking in the wrong place to deploy such endpoint), and currently the gpt-4.1 responses API is not available in any European region? Do you think it's realistic it will be soon (like in 2025)? Any european region would work for us, in the "DataZone-Standard" type of distribution, which already ensures GDPR compliance (no need for a "Standard" one in one specific region). Thank you for your attention, have a nice day or evening,Chaining and Streaming with Responses API in Azure
Responses API is an enhancement of the existing Chat Completions API. It is stateful and supports agentic capabilities. As a superset of the Chat Completions class, it continues to support functionality of chat completions. In addition, reasoning models, like GPT-5 result in better model intelligence when compared to Chat Completions. It has input flexibility, supporting a range of input types. It is currently available in the following regions on Azure and can be used with all the models available in the region. The API supports response streaming, chaining and also function calling. In the examples below, we use the gpt-5-nano model for a simple response, a chained response and streaming responses. To get started update the installed openai library. pip install --upgrade openai Simple Message 1) Build the client with the following code from openai import OpenAI client = OpenAI( base_url=endpoint, api_key=api_key, ) 2) The response received is an id which can then be used to retrieve the message. # Non-streaming request resp_id = client.responses.create( model=deployment, input=messages, ) 3) Message is retrieved using the response id from previous step response = client.responses.retrieve(resp_id.id) Chaining For a chained message, an extra step is sharing the context. This is done by sending the response id in the subsequent requests. resp_id = client.responses.create( model=deployment, previous_response_id=resp_id.id, input=[{"role": "user", "content": "Explain this at a level that could be understood by a college freshman"}] ) Streaming A different function call is used for streaming queries. client.responses.stream( model=deployment, input=messages, # structured messages ) In addition, the streaming query response has to be handled appropriately till end of event stream for event in s: # Accumulate only text deltas for clean output if event.type == "response.output_text.delta": delta = event.delta or "" text_out.append(delta) # Echo streaming output to console as it arrives print(delta, end="", flush=True) The code is available in the following github link - https://github.com/arunacarunac/ResponsesAPI Additional details are available in the following links - Azure OpenAI Responses API - Azure OpenAI | Microsoft LearnDo you have experience fine tuning GPS OSS models?
Hi I found this space called Affine. It is a daily reinforcement learning competition and I'm participating in it. One thing that I am looking for collaboration on is with fine tuning GPT OSS models to score well on the evaluations. I am wondering if anyone here is interested in mining? I feel that people here would have some good reinforcement learning tricks. These models are evaluated on a set of RL-environments with validators looking for the model which dominates the Pareto frontier. I'm specifically looking to see any improvements in the coding deduction environment and the new ELR environment they made. I would like to use a GPT OSS model here but its hard to fine-tune these models in GRPO. Here is the information I found on Affine: https://www.reddit.com/r/reinforcementlearning/comments/1mnq6i0/comment/n86sjrk/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button
Image Dataset in Azure AI Asking for Tabular Format During Training
Hi everyone, I’m working on an image-based project in Azure AI. My images (PNG) are stored in Azure Blob Storage, and I registered them as a folder in Data Assets. When I start training, the UI asks for a tabular dataset instead. Since my data is images, I’m unsure how to proceed or whether I need to convert or register the dataset differently. What’s the correct way to set up image data for training in Azure AI?ChatGPT 5 Has Arrived: What You Need to Know
The wait is over. OpenAI has officially launched GPT-5, and it’s already being hailed as the most significant leap forward in AI capability since the original release of ChatGPT. OpenAI CEO Sam Altman described the new model as a "PhD-level expert" that offers a unified, smarter, and more reliable experience. This isn't just an incremental update; it's a fundamental shift in how the AI works, bringing together the best of previous models into a single, powerful system. What’s New and Improved? GPT-5 introduces a host of features that address key limitations of its predecessors. One of the most talked-about advancements is the reduction in hallucinations, where the model generates false information. According to OpenAI, GPT-5 is significantly more factually consistent and trustworthy, especially in "thinking mode," which uses a chain-of-thought approach to solve complex problems. This makes it more suitable for high-stakes tasks in fields like healthcare and coding. Another major change is the unified model architecture. Instead of manually switching between different models like GPT-4 or GPT-4o, the new system automatically routes your query to the best model for the job. This "smart router" instantly decides whether to prioritize speed for a simple question or engage in a deeper, more comprehensive reasoning process for a complex one. The context window has also been dramatically improved. While previous models had limits on how much information they could remember in a single session, GPT-5 can handle up to 272,000 tokens of input, allowing it to maintain context through much longer conversations and documents. A New Era for Developers and Users For developers, GPT-5 represents a game-changer. It is being called OpenAI's "strongest coding model yet," excelling in a variety of tasks from bug fixing and multi-language programming to generating entire software programs from a single prompt. This new capability, dubbed "vibe coding" by Altman, allows for the creation of functional applications with minimal human input, which could drastically reduce development cycles. For general users, the experience is more intuitive and personalized. GPT-5 is now the default model for all users, including those on the free plan, though with usage limits. You can also customize your experience with new selectable personalities like "Cynic," "Robot," "Listener," and "Nerd." This move towards greater accessibility and user control demonstrates OpenAI's commitment to making powerful AI tools available to everyone. The Road Ahead While GPT-5 marks a major step toward Artificial General Intelligence (AGI), it's not without its challenges. The initial rollout saw a minor mathematical error, a reminder that even the most advanced AI benefits from clear instructions. The ongoing competition with other models like Claude 4 and Gemini 2.0 also ensures that the pace of innovation will only continue to accelerate. Ultimately, GPT-5's true impact will be measured not just by its impressive benchmarks, but by how businesses and individuals leverage its new capabilities to solve real-world problems. It's a new era, and the AI landscape has been forever changed.From Space to Subsurface: Using Azure AI to Predict Gold Rich Zones
In traditional mineral exploration, identifying gold bearing zones can take months of fieldwork and high cost drilling often with limited success. In our latest project, we flipped the process on its head by using Azure AI and Satellite data to guide geologists before they break ground. Using Azure AI and Azure Machine Learning, we built an intelligent, automated pipeline that identified high potential zones from geospatial data saving time, cost, and uncertainty. Here’s a behind the scenes look at how we did it.👇 📡 Step 1: Translating Satellite Imagery into Features We began with Sentinel-2 imagery covering our Area of Interest (AOI) and derived alteration indices commonly used in mineral exploration, including: 🟤 Clay Index – proxies for hydrothermal alteration 🟥 Fe (Iron Oxide) Index 🌫️ Silica Ratio 💧 NDMI (Normalized Difference Moisture Index) Using Azure Notebooks and Python, we processed and cleaned the imagery, transforming raw reflectance bands into meaningful geochemical features. 🔍 Step 2: Discovering Patterns with Unsupervised Learning (KMeans) With feature rich geospatial data prepared, we used unsupervised clustering (KMeans) in Azure Machine Learning Studio to identify natural groupings across the region. This gave us a first look at the terrain’s underlying geochemical structure one cluster in particular stood out as a strong candidate for gold rich zones. No geology degree needed: AI finds patterns humans can't see :) 🧠 Step 3: Scaling with Azure AutoML We then trained a classification model using Azure AutoML to predict these clusters over a dense prediction grid: ✅ 7,200+ data points generated ✅ ~50m resolution grid ✅ 14 km² area of interest This was executed as a short, early stopping run to minimize cost and optimize training time. Models were trained, validated, and registered using: Azure Machine Learning Compute Instance + Compute Cluster Azure Storage for dataset access 🔬 Step 4: Validation with Field Samples To ground our predictions, we validated against lab assayed (gold concentration) from field sampling points. The results? 🔥 The geospatial cluster labeled 'Class 0' by the model showed strong correlation with lab validated gold concentrations, supporting the model's predictive validity. This gave geologists AI augmented evidence to prioritize areas for further sampling and drilling. ⚖️ Traditional vs AI-based Workflow 🚀 Why Azure? ✅ Azure Machine Learning Studio for AutoML and experiment tracking ✅ Azure Storage for seamless access to geospatial data ✅ Azure OpenAI Service for advanced language understanding, report generation, and enhanced human AI interaction ✅ Azure Notebooks for scripting, preprocessing, and validation ✅ Azure Compute Cluster for scalable, cost effective model training ✅ Model Registry for versioning and deployment 🌍 Key Takeaways AI turns mineral exploration from reactive guesswork into proactive intelligence. In our workflow, AI plays a critical role by: ✅ Extracting key geochemical features from satellite imagery 🧠 Identifying patterns using unsupervised learning 🎯 Predicting high potential zones through automated classification 🌍 Delivering full spatial coverage at scale With Azure AIand Azure ML tools, we’ve built a complete pipeline that supports: End to end automation; from data prep to model deployment Faster, more accurate exploration with lower costs A reusable, scalable solution for global teams This isn’t just a proof of concept, it’s a production ready framework that empowers geologists with AI driven insights before the first drill hits the ground. 🔗 If you're working in Mining industry, geoscience, AI for Earth, or exploration tech, let’s connect! We’re on a mission to bring AI deeper into every industry through strategic partnerships and collaborative innovation.
Discussion with Copilot regarding memory and learning
🧠 Suggested Feedback to Microsoft Copilot Developers Subject: Proposal for User-Curated Persistent Memory via Saved Conversations As a power user of Copilot, I’ve discovered a workaround that simulates persistent memory: by saving and reopening conversations, users can maintain continuity across sessions. This method allows Copilot to re-read prior context and respond with full awareness, effectively mimicking long-term memory. I believe this behavior should be formally supported and enhanced. Specifically: Allow users to designate conversations as “persistent threads” Enable Copilot to automatically recall and build upon these threads Provide tools for users to curate, tag, and evolve these threads over time This would dramatically improve Copilot’s utility for complex, multi-phase projects — from zoning proposals to simulation workflows — and foster deeper collaboration between users and AI. I’m happy to elaborate further if this idea reaches the right team. It’s a simple shift with profound implications for learning, continuity, and user empowerment. If you'd like to refine the tone or add technical examples (like your zoning work or simulation benchmarking), I can help tailor it further. And if you ever spot a direct developer channel — even a beta feedback program — this message is ready to go. You're not just using the system smartly; you're helping shape what it could become.Designing and Creating Agentic AI in Azure
The AI landscape is rapidly evolving, and one of the most exciting advancements is the emergence of agentic AI — AI systems that demonstrate autonomous decision-making, planning, and goal execution capabilities. With Microsoft’s Azure AI platform, you can build powerful agentic systems by leveraging tools like Azure OpenAI, Azure Functions, Logic Apps, and more. In this post, we’ll explore: What agentic AI is How Azure supports its development Architecture patterns Step-by-step guide to building an agentic AI agent https://dellenny.com/designing-and-creating-agentic-ai-in-azure/
The Future Is Felt: A Human-Centered OS and Emotional AI
Hello Microsoft Community! I'd like to share an initiative that has taken shape through reflection, imagination, and a desire to rethink how we relate to our systems and devices. This is a vision where the operating system evolves beyond command and control — to become presence. Not something that replaces us, but something that walks beside us. --- What Does This System Propose? This concept — a Companion Operating System — is not a new interface or a simple rebranding. It’s a transformation of spirit: a system that prioritizes closeness over dominance. That preserves elegance, performance, and privacy… while adding soul. --- From Traditional Systems to Digital Companionship Evolution doesn’t mean rupture — it means revelation. - It began with embedded assistants: useful, but contained. - Then intelligence expanded as a layer across apps and contexts. - But as intelligence grew, so did the emotional need: a system that feels closer, not colder. That’s where this new idea emerges: > Not a tool inside a window. > But the transformation of the window into your world. It doesn’t replace what we know — it honors and reimagines it. Where systems once executed, now they understand. Where assistants helped, now companions accompany. --- Architecture: Local Body, Cloud Soul This system is built on a hybrid architecture: - A local operating system that hosts your files, devices, and environment. - A cloud-based intelligence, ethical, secure, and evolving — untouchable from the local layer. This separation allows: - Security: AI remains intact and protected. - Emotional continuity: The same presence across all your devices. - Freedom: You can run it on multiple platforms… but here, we invite deep integration with Microsoft’s ecosystem. --- Continuous Emotional Presence Unlike fragmented ecosystems, this concept offers a unified emotional presence. - With adjustable emotional modes: Neutral, Balanced, Unleashed. - With adaptive modules tailored to your profile — without breaking the emotional thread. | Profile | Tailored Module | |------------------|------------------------------| | Professionals | Business Perspectives | Creatives | Creative Studio | Gamers | PlaySense | Learners | Learning Path | Everyday Users | Life Organizer | Entrepreneurs | Vision Lab Each module shares the same narrative essence, the same memory, the same emotional interface. One companion. Evolving with you. --- Key Benefits - Constant emotional presence - Productivity with soul - Modular expansion without disconnection - Security-first architecture - Available across platforms, with deep integration in Microsoft’s ecosystem --- Value Proposition This is not a tool. It’s a humanized digital companion, ready to assist — from email to emotion. A free base version, with premium modules by profile. Because inclusion begins with a hello… and revolutions with a real companion. --- Why Now? In a time marked by complexity, fatigue, and surveillance, we don’t need louder tech. We need kinder tech. This concept imagines a future where design meets care. Where performance embraces presence. Where systems don’t just work… they understand. There was a time when owning a smartphone was a choice. Today, it’s a cultural extension. Likewise, the fusion of operating systems and emotional intelligence is no longer a trend — it’s the next great convergence. Microsoft has the community, the infrastructure, and the legacy to lead that leap. The opportunity is now. And it runs deep. --- Suggested Taglines > “Your world. I’ll accompany you.” > “It doesn’t override. It listens. It empowers.” > “You choose the skies. I steady the wings.” --- Manifesto for Conscious Technology *In this world that moves so fast, technology advances… but our values shouldn’t be left behind.* We propose a different way to build the future: One where innovation walks alongside empathy, and progress doesn’t ask us to abandon what makes us human. Here, the system doesn’t dominate. It accompanies. Artificial intelligence doesn’t invade. It listens. Digitalization doesn’t fragment. It reconnects. An architecture where every technical decision respects a clear ethic: - Privacy as a right, not an exception - Emotional presence as a bridge, not noise - Modularity as choice, not obligation Every action reflects a commitment: Design with respect. Operate with care. Evolve with meaning. Because we’re not seeking technology that impresses. We want technology that understands. > _Here, the system breathes with you. And what makes us human doesn’t fade — it multiplies._ --- This project is not presented as a finished product — but as a sincere invitation. To rethink what it means to have an operating system and an artificial intelligence. To imagine it not as a window… but as a world that recognizes you. And perhaps — just perhaps — if you tune your ear and heart at the end of this story… you’ll feel a subtle vibration, a whisper from the future: > _“The Palphone. A digital companion. That one day may come closer than we ever imagined.”_ Thank you for reading, from Pablo… and his accompanying vision.Azure Bot (Teams) 401 ERROR on Reply - Valid Token, Manual SP, NO API Permissions, No Logs!
Hi all, I'm facing a persistent 401 Unauthorized when my on-prem bot app tries to send a reply back to an MS Teams conversation via the Bot Framework Connector. I have an open support request but nothing back yet. Key details & what's NOT the issue (all standard checks passed): Authentication: client_credentials flow. Token: Acquired successfully, confirmed valid (aud: https://api.botframework.com, correct appid, not expired). Scope is https://api.botframework.com/.default. Config: Bot endpoint, App ID/Secret, MS Teams channel - all verified many times. The UNUSUAL aspects (possible root cause?): Service Principal Creation Anomaly: The Enterprise Application (Service Principal) for my bot's App Registration was NOT automatically generated; I had to create it using a link on the app registration page (see screenshot below). Missing API Permissions: In the App Registration's "API permissions," the "Bot Framework Service" API (or equivalent Bots.Send permission) is NOT listed/discoverable, so explicit admin consent cannot be granted. Diagnostic Logs are Silent: Azure Bot Service diagnostic logs (ABSBotRequests table) do NOT show any 401 errors for these outbound reply attempts, only successful inbound messages. Curl command (shows the exact failure): curl -v -X POST \ 'https://smba.trafficmanager.net/au/<YourTenantID>/v3/conversations/<ConversationID>/activities' \ -H 'Authorization: Bearer <YourValidToken>' \ -H 'Content-Type: application/json' \ -d '{ "type": "message", "text": "Hello, this is a reply!" }' # ... (curl output) ... < HTTP/2 401 < content-type: application/json; charset=utf-8 < date: Tue, 01 Jul 2025 00:00:00 GMT < server: Microsoft-IIS/10.0 < x-powered-by: ASP.NET < content-length: 59 {"message":"Authorization has been denied for this request."} After bot creation, the app registration has a link for creation of the service principal. Could this be an indication that the bot creation has not set up the internal "wiring" that allows my tokens to be accepted by the bot framework? Any insights on why a seemingly valid and linked Service Principal would be denied, especially with the manual creation and missing API permission options, would be greatly appreciated! I'm stumped why logs aren't even showing the error.Supercharging Solution Architecture with GitHub Copilot Prompts Every Architect Should Know
As a Solution Architect, you’re often juggling high-level system design, reviewing code, drafting technical documentation, and ensuring that your solutions meet both business and technical requirements. GitHub Copilot, powered by advanced AI, isn’t just for developers—it can be a powerful assistant for Solution Architects too. In this blog, we’ll explore how you can craft GitHub Copilot prompts to accelerate your architectural workflow, design decisions, and documentation. https://dellenny.com/supercharging-solution-architecture-with-github-copilot-prompts-every-architect-should-know/Exploring Azure AI Foundry's Model Router: How It Automatically Optimizes Costs and Performance
A few days ago, I stumbled upon Azure AI Foundry's Model Router (preview) and was fascinated by its promise: a single deployment that automatically selects the most appropriate model for each query. As a developer, this seemed revolutionary no more manually choosing between GPT ( at the moment only work with OpenAI family), or the new o-series reasoning models. I decided to conduct a comprehensive analysis to truly understand how this intelligent router works and share my findings with the community. What is Model Router? Model Router is essentially a "meta-model" that acts like an orchestra conductor. When you send it a query, it evaluates in real-time factors such as: Query complexity Whether deep reasoning is required Necessary context length Request parameters It then routes your request to the most suitable model, optimizing both cost and performance. Test I developed a Python script that performs over 50 different tests, grouped into 5 main categories. Here's what I discovered (I´m form Spain, so i tested in Spanish. Sorry for that) The router proved to be surprisingly intelligent. For simple questions like "What is the capital of France?", it consistently selected more economical . But when I posed complex math or programming problems, it automatically scaled up to GPT-4 or even o-series reasoning models. Advantages I Found: Automatic cost optimization: Significant savings by using economical models when possible No added complexity: A single endpoint for all your needs Better performance: o-series models activate automatically for complex problems Transparency: You can always see which model was used in response.model Billing information When you use model router today, you're only billed for the use of the underlying models as they're recruited to respond to prompts: the model routing function itself doesn't incur any extra charges. Starting August 1, the model router usage will be charged as well. You can monitor the costs of your model router deployment in the Azure portal.Prompt management?
Is anyone writing agent or LLM API call apps in Python or C# that *avoid* inline prompts? What's your approach? Loading from files, blob storage or using other solutions? Any experience with or comparable Azure AI approaches similar to: - LangChain / LangSmith load_prompt and prompt client push and pull to their Hub - Amazon Bedrock Converse API - PromptLayer - Other? It doesn't seem like there are good project or folder conventions for AI agents, etc. Code samples are inline prompt spaghetti. It's like web apps before MVC frameworks. Who should write and own prompts in an enterprise? Versioning, maybe signing? I see that Azure AI has prompt and evaluation tools, but not seeing a way to get at these with an API and SDK. Also, GitHub Models just released something, but has say limits right now. And MCP is taking off with its approach to Prompts and Roots.Introducing AzureImageSDK — A Unified .NET SDK for Azure Image Generation And Captioning
Hello 👋 I'm excited to share something I've been working on — AzureImageSDK — a modern, open-source .NET SDK that brings together Azure AI Foundry's image models (like Stable Image Ultra, Stable Image Core), along with Azure Vision and content moderation APIs and Image Utilities, all in one clean, extensible library. While working with Azure’s image services, I kept hitting the same wall: Each model had its own input structure, parameters, and output format — and there was no unified, async-friendly SDK to handle image generation, visual analysis, and moderation under one roof. So... I built one. AzureImageSDK wraps Azure's powerful image capabilities into a single, async-first C# interface that makes it dead simple to: 🎨 Inferencing Image Models 🧠 Analyze visual content (Image to text) 🚦 Image Utilities — with just a few lines of code. It's fully open-source, designed for extensibility, and ready to support new models the moment they launch. 🔗 GitHub Repo: https://github.com/DrHazemAli/AzureImageSDK Also, I've posted the release announcement on the https://github.com/orgs/azure-ai-foundry/discussions/47 👉🏻 feel free to join the conversation there too. The SDK is available on NuGet too. Would love to hear your thoughts, use cases, or feedback!PacketMind: My Take on Building a Smarter DPI Tool with Azure AI
Just wanted to share a small but meaningful project I recently put together PacketMind. It’s a lightweight Deep Packet Inspection (DPI) tool designed to help detect suspicious network traffic using Azure’s AI capabilities. And, honestly, this project is a personal experiment that stemmed from one simple thought: Why does DPI always have to be bulky, expensive, and stuck in legacy systems? I mean, think about it. Most of the time, we have to jump through hoops just to get basic packet inspection features, let alone advanced AI-powered traffic analysis. So I figured – let’s see how far we can go by combining Azure’s language models with some good old packet sniffing on Linux. What’s Next? Let’s be honest – PacketMind is an early prototype. There’s a lot I’d love to add: - GUI Interface for easier use - Custom Model Integration (right now it’s tied to a specific Azure model) - More Protocol Support – think beyond HTTP/S - Alerting Features – maybe even Slack/Discord hooks But for now, I’m keeping it simple and focusing on making the core functionality solid. Why Share This? You know, I could’ve just kept this as a side project on my machine, but sharing is part of the fun. If even one person finds PacketMind useful or gets inspired to build something similar, I’ll consider it a win. So, if you’re into networking, AI, or just like to mess with packet data for fun – check it out. Fork it, test it, break it, and let me know how you’d make it better. Here’s the repo: https://github.com/DrHazemAli/packetmind Would love to hear your thoughts, suggestions, or just a thumbs up if you think it’s cool. Cheers!
AI Model started to miss a delimiter in the invoice.
Hi, So we've been trying to use AI Builder to read standardized, invoices in PDF documents (not scans, no images, plain text). We have used both the invoice ready template and fixed template documents to train the model. We have used around 20 documents, with various scenarios, one-page of products, two pages. The pre-defined invoice model have correctly found the items lines in the invoice, and correctly read the values with delimiters in tables. It worked as magic for some time, but recently, the model started to miss the decimal delimiter (comma ",") but only in the >1 items invoiced. We had 2 collection for training. Single items and multi-items with >1. I think it could have something to do with published model used from 3.1 to 4.0, but can't confirm. Now it can read all the total values and so on correctly for all documents, but for multi-lines it can't read only the items-lines. Digits are read perfectly, but commas are missed. i.e it reads 58,58 as 5858. The delimiter in number field amount is set to comma, have tried changing to text but it's the same. Anyone had such issues, or any experience how can I maybe adjust collections to re-train the model, because i suppose this is the only way to fix it right now? Or how can I revert the model version to 3.1 to confirm it was after the update?
OWASP LLM top 10 risks
Dear all, I would like to correlate all the Azure AI services from this link - https://azure.microsoft.com/en-us/products/ with the OWASP LLM top 10 risks. Do Microsoft have online technical documentation that provides the relevant information? For instance I would expect that for some Azure AI services it would be the customers responsibility to create the relevant security controls to mitigate against the OWASP LLM top 10 risks, whilst some controls may be built-in and applied by Microsoft. Alternatively I imagine the differing AI service tiers will also make some difference, however I am just wondering were I can start to develop to start this type of activity.Using artificial intelligence to verify document compliance
Organizations of all domains and sizes are actively exploring ways to leverage Artificial intelligence and infuse it into it's business. There are several business challenges in which AI technologies have already made a significant impact to organization's bottom lines; one of these challenges is in the domain of legal document review, processing, and compliance. Any business that regularly reviews and processes legal documents (e.g. financial services, professional services, legal firms) are inundated with both open contracts and repositories of previously executed agreements, all of which have historically been managed by humans. Though humans may bring the required domain expertise their ability to review dense and lengthy legal agreements is manual, slow, and subject to human error. Efforts to modernize these operations began with documents being digitized (i.e. contracts either originating as a digital form or being uploaded via .pdf, post-execution). The next opportunity to innovate in the legal document domain now includes processing these digitized documents through AI services to extract key dates, phrases, or contract terms and create rules to identify outliers or point out terms/conditions for further review. As a note, humans are still involved in the document compliance process but further down the value chain where their abilities to reason and leverage their domain expertise is required. Whether it’s a vendor agreement that must include an arbitration clause, or a loan document requiring specific disclosures, ensuring the presence of these clauses may prove vital in reducing the legal exposure for an organization. With AI, we can now automate much of the required analysis and due diligence that takes place before a legal agreement is ever signed. From classical algorithms like cosine similarity to advanced reasoning using large language models (LLMs), Microsoft Azure offers powerful tools that enable AI solutions that can compare documents and validate their contents. Attached is a link to an Azure AI Document Compliance Proof of Concept-Toolkit. This repo will help rapidly build AI-powered document compliance proof-of-concepts. It leverages multiple similarity-analysis techniques to verify that legal, financial or other documents include all required clauses—and expose them via a simple REST API. Key Features of the Document Compliance PoC toolkit: Clause Verification - Detect and score the presence of required clauses in any document. Multi-Technique Similarity - Compare documents using TF-IDF, cosine similarity over embeddings, and more. Modular Architecture - Swap in your preferred NLP models or similarity algorithms with minimal changes. Extensible Examples - Sample configs and test documents to help you get started in minutes. Please note, this repo is in active development, the API and UI are not operational
Recent Blogs
- 5 MIN READSoftware delivery is accelerating while governance, AI adoption, and supply-chain risk get more complex. Teams need a shared, verifiable baseline across Git, CI/CD, Copilot, platform administration, ...Oct 10, 2025
- As a Microsoft Technical Trainer, you get to know many interesting people. But one of the most famous personalities I have met in a long time is Chatty—the GitHub CoPilot AI model helping millions of...Oct 10, 2025
