Agents vs project type vs capability host

[haughty zealot]

I understand there is an ongoing transition between AI Project (the projects which use AI Hub) and Foundry projects - the projects which only use foundry.

The recommendation is to focus on the Foundry projects, correct ?

All my tests to use custom azure functions, MCPs, etc, were on Foundry projects and were failing in AI projects. Is this correct, or should I push to use the same resources in AI projects as well?

The problem: The Foundry projects only allow the creation of agents if there is a capability host present. If there is no capability host, I can't create an agent - the option doesn't appear in the UI.

The capability host requires a cosmos db to work. Cosmos db is expensive - at least for simple tests, it consumes a lot. The AI projects consume much less.

I tried to use cosmos db serverless, but the foundry project doesn't allow it. The agent creation is disabled.

Is there a way to deploy a foundry project with support to agent creation but using less expensive resources than cosmos? A simple cosmosdb serverless could solve the problem.

I also sent the question in github, but I already sent about 5 questions there and got no reply 🙁

[silk adder]

I found that using the m365 agents SDK in vs code to create and deploy is an easier solution. Especially if you don't have full azure admin permissions. With the SDK I was able to configure what I wanted to use for different workflows, such as using blob storage instead of cosmos db. I found the azure project templates frustrating and the agent creation in foundry would only use openai models. Once I used the m365 agent sdk and added semantic kernel / agent framework, it unlocked all the limitations I found in ai foundry. Multiple workflows, multiple agents with any model from a single resource hub all by vibe coding with Claude Sonnet.