Foundry agents with tools
Agents built in Microsoft Foundry (formerly Azure AI Foundry) that call functions, search indexes and enterprise systems on behalf of users.
SERVICES / CLOUD PLATFORMS
Evaluate agents built on Microsoft Azure.
Independent testing and evaluation for AI agents built with Microsoft Foundry, Azure OpenAI and Copilot Studio, combining native evaluators with your business-specific checks.
Common Azure agent patterns
Copilot Studio agents
Business-built agents and copilots connected to Microsoft 365, Dataverse and line-of-business data, often created outside the core engineering team.
Azure OpenAI applications
Custom agents and orchestration code calling Azure OpenAI models in Microsoft Foundry, frequently combined with Azure AI Search for retrieval.
Quality issues we look for
Misread intent
The agent solves a different problem from the one the user asked about, especially when requests are short, ambiguous or change mid-conversation.
Incorrect tool calls
The right tool with the wrong arguments, unnecessary calls, or missing calls that leave a workflow incomplete.
Instructions drift
Responses that ignore system instructions, required steps or company policy as conversations grow longer.
What we assess
Intent and task adherence
Whether the agent understood the request and completed the task within its instructions, tested on your real scenarios.
Tool-call accuracy and outcome
Whether tools were chosen and called correctly, and whether the resulting records match what should have happened.
Grounded responses
Whether answers drawn from your search indexes and documents are relevant, complete and faithful to the source.
Working with native Azure evaluation
Built-in agent evaluators
Microsoft Foundry evaluation tooling includes agent-focused evaluators for intent resolution, tool-call accuracy and task adherence. We configure them with your data where they fit.
Custom evaluators for your rules
Business-specific checks and rubrics fill the gaps generic evaluators can't cover, such as eligibility rules or required disclosures.
Preview features handled carefully
Some evaluators may be in preview. We confirm status and limitations in discovery and avoid relying on unstable features for release gates.
Plan the engagement
Your inputs
A test project or environment, agreed access, deployment and version details, business rules, example tasks and a domain reviewer.
Your deliverables
An agreed benchmark, prioritized findings, reusable evaluation assets and recommendations for the next iteration.
Support after handover
We can retest fixes, extend coverage to new Azure workflows or move to recurring Managed Evaluations as your agent changes. Platform usage and licenses stay in your account.
Make your next decision with evidence.
Start with one workflow, clear success criteria, and an evaluation plan your team can use.