Build Production AI Agents
Not Infrastructure

Huge respect to the team at Vstorm for pushing this forward. That's the kind of runtime thinking we need if we want agents in production, not just demos.

The CLI angle is underrated. The hard part isn't generating code — it's standardizing how teams spin things up so auth, streaming, and observability aren't afterthoughts. Tooling like this matters because it encodes good defaults before entropy shows up.

Great overview on Pydantic Deep Agents. Thanks Vstorm!

Setting up the production plumbing often eats up time before you even touch the AI logic. I've seen teams lose momentum getting stuck on this infrastructure glue. This CLI Generator looks like a solid way to fast-track deployment.

Our team hasn't tried the full library yet but have been finding a lot of value in a couple of the sub-libraries, many thanks!

This is how agents move from demos to products. The scaffolding matters.

I saw your post regarding deep agents and I'm super interested! I was working on history processor and summarization, enabling the agent to self-compress when we give it a warning about its own context window size.

Really appreciate the work done here: it was thoughtful of you to include the admin panel — that's been a big hurdle whenever I work with FastAPI.

We're looking to expand our agentic capabilities, and your team's work seems very aligned with where we're headed. We could leverage the existing pydantic-ai-backend and extend it to other remote execution environments.

Perfect. Makes scaffolding e2e agent projects that much easier.

Excellent end to end Template for AI/LLM Applications.

I made a PR for pydantic todos — some instruction fixes, also make todo ids show in the prompt so you can make one less tool call. I will take care of the subagents one in a bit.

Huge respect to the team at Vstorm for pushing this forward. That's the kind of runtime thinking we need if we want agents in production, not just demos.

The CLI angle is underrated. The hard part isn't generating code — it's standardizing how teams spin things up so auth, streaming, and observability aren't afterthoughts. Tooling like this matters because it encodes good defaults before entropy shows up.

Great overview on Pydantic Deep Agents. Thanks Vstorm!

Setting up the production plumbing often eats up time before you even touch the AI logic. I've seen teams lose momentum getting stuck on this infrastructure glue. This CLI Generator looks like a solid way to fast-track deployment.

Our team hasn't tried the full library yet but have been finding a lot of value in a couple of the sub-libraries, many thanks!

This is how agents move from demos to products. The scaffolding matters.

I saw your post regarding deep agents and I'm super interested! I was working on history processor and summarization, enabling the agent to self-compress when we give it a warning about its own context window size.

Really appreciate the work done here: it was thoughtful of you to include the admin panel — that's been a big hurdle whenever I work with FastAPI.

We're looking to expand our agentic capabilities, and your team's work seems very aligned with where we're headed. We could leverage the existing pydantic-ai-backend and extend it to other remote execution environments.

Perfect. Makes scaffolding e2e agent projects that much easier.

Excellent end to end Template for AI/LLM Applications.

I made a PR for pydantic todos — some instruction fixes, also make todo ids show in the prompt so you can make one less tool call. I will take care of the subagents one in a bit.