How Microsoft Foundry and Elastic Agent Builder in Microsoft Marketplace are shaping enterprise AI

In this guest blog post, Brian Bergholm, Principal Product Marketing Manager at Elastic, covers the evolution of agentic AI and how enterprises can leverage context engineering via Microsoft Foundry and Elastic Agent Builder in the Microsoft Marketplace to operationalize agentic systems at scale.

For the past few years, AI conversations have focused heavily on large language models (LLMs): bigger models, faster models, and multimodal models. But a quieter (and arguably more important) shift is underway. The real differentiation is moving above the model layer, toward how systems understand context and take action.

The shift is particularly evident in the evolution from simple, conversational chat interfaces to sophisticated, independent AI agents. These agents are no longer just responding to prompts; they are executing specific tasks, making decisions, and managing complex workflows with their effectiveness deeply informed by advanced context engineering.

This is where context engineering and agentic AI come together and where platforms like Elastic Agent Builder and Microsoft Foundry (formerly known as Azure AI Foundry) are shaping what enterprise AI looks like in practice.

The end of prompt-centric AI

Prompt engineering was a necessary first step. It taught us how sensitive large language models are to instructions, framing, and examples. But prompt-centric AI doesn’t naturally scale well in the enterprise. Why? Prompts lack awareness of real-time data, rely on humans to manually provide context, and don’t translate easily into repeatable systems.

Enter context engineering. Context engineering is the discipline of systematically constructing the information environment AI operates within. Instead of treating context as an afterthought, it becomes a first-class design principle by including:

• Relevant data (logs, metrics, tickets, and documents)
• User and system state
• Goals and constraints
• Policies, guardrails, and domain knowledge

The insight driving this trend is simple but powerful: AI output quality is bounded by the quality and relevance of the context it receives.

As enterprises move from demos to production, context engineering is replacing clever prompts as the primary lever for improving outcomes.

Once context is engineered effectively, the next evolution is inevitable: AI systems that can act. Agentic AI systems differ from traditional assistants in a few critical ways:

• They operate toward a goal, not just a query.
• They can plan multistep actions.
• They use tools and data sources.
• They adapt based on outcomes.

This matters because enterprise problems are rarely single-turn questions. They are workflows that can:

• Investigate an incident
• Diagnose a security alert
• Optimize system performance
• Assist a customer through resolution

Agentic AI represents a shift from AI as an answer engine to AI as an operational participant.

Why are search and observability foundational to agents?

There’s a hard truth about agentic AI: Reasoning without reliable data is just speculation. High-impact agents need fast access to relevant information, the ability to explore and refine queries, and confidence that their results reflect reality, not stale knowledge. Retrieval augmented generation (RAG) provides the right context for every AI answer, ensuring its accuracy.

This is where Elastic Agent Builder becomes strategically important. Elastic Agent Builder allows teams to create AI agents that search and retrieve context from general or proprietary data; ground reasoning in logs, metrics, traces, security events, and content; and continuously refresh context as systems change.

In practice, this means agents can investigate incidents using real-time observability data, correlate signals across systems, and explain why something is happening, not just what happened.

In an agentic world, search is not a feature; it’s a cognitive function.

Microsoft Foundry: Operationalizing agentic systems at scale

While Elastic excels at context and real-time insight, enterprises still need a way to orchestrate, govern, and scale AI systems. This is the role of Microsoft Foundry. Microsoft Foundry provides:

• A platform for managing models, agents, and tools
• Built-in security, compliance, and governance
• Integration with enterprise workflows and applications

In other words, Foundry addresses the question “How do we safely run agentic AI in production?”

When combined with Elastic’s ability to supply high-fidelity, real-time context, Foundry orchestrates the agent lifecycle, permitting agents to reason, act, and adapt with guardrails in place.

The convergence: Systems that understand, decide, and act

What we’re seeing is not just a tooling trend; it’s an architectural shift.

Elastic Agent Builder and Microsoft Foundry sit at complementary layers of this stack:

• Elastic ensures agents stay grounded in truth
• Foundry ensures agents operate responsibly and at scale

Together, they enable a new class of enterprise AI: systems that are not just intelligent, but also situationally aware and operationally useful.

Looking ahead: Context is the competitive advantage

As models continue to commoditize, the winners won’t be those with the biggest parameters. Instead, they’ll be the ones who:

• Engineer context deliberately
• Embed AI into real workflows
• Trust agents with meaningful responsibility
• Govern those agents effectively

Context engineering and agentic AI are not future concepts. They are already reshaping how work gets done. The organizations that embrace this shift — grounded in platforms like Elastic and Microsoft Foundry — will move faster, respond smarter, and operate with a level of intelligence static systems simply can’t match.

Learn more about context engineering and how autonomous systems are changing cybersecurity. Or, start a seven-day free trial of Elastic on Microsoft Marketplace.