What is the difference between Foundry and Azure OpenAI?
Azure OpenAI gives you API access to OpenAI models. Microsoft Foundry is the wider environment where you orchestrate those models alongside other tools, manage data connections and govern the application's lifecycle. You build in Foundry, and Azure OpenAI is one of the model services you build with.
Is Microsoft Foundry free?
No. Microsoft charges no platform fee, but you pay as you go for model tokens, hosted-agent compute and each tool the agent uses; a hosted agent costs $0.0994 per vCPU-hour in East US (checked 4 October 2026). You can explore at low cost, and production workloads cost what they use. We estimate the running cost of your workflow during scoping.
How does Foundry handle data privacy and security?
Foundry runs inside your Azure tenant under Microsoft's enterprise security model, and your data is not used to train public models. You get data residency controls, role-based access and private connections to internal sources. We add a context layer so each agent reads only the fields its task needs.
Foundry or Copilot Studio?
Copilot Studio lets business users build agents inside Microsoft 365 themselves. Foundry is built for developers and gives full control over models, code and hosting. If the agent needs custom logic, several models or strict security sign-off, Foundry is the better base.
Can you take over a Foundry project another team started?
Yes. We start with a technical assessment of what is built, then fix what it finds, harden the agent for production and run it in weekly sprints. See AI managed services.
Where should we start?
With the workflow that has the most volume and the clearest rules. If you have several candidates, the AI Audit ranks them over 4 to 8 weeks and ends in a 90-day roadmap, for $10,000 to $100,000 depending on scope, with the fee credited toward your first build.
