# PydanticAI Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/pydanticai
> Markdown URL: https://aitinkerers.org/technologies/pydanticai.md
> Technology record last updated: 2026-03-01T04:48:45Z
> Generated: 2026-09-20T14:48:28Z

A Python agent framework for building production-grade, type-safe Generative AI applications with validated, structured outputs.

PydanticAI is the Python agent framework from the Pydantic team, designed to bring FastAPI's ergonomic, type-safe development experience to Generative AI. It leverages Pydantic’s core data validation to ensure Large Language Model (LLM) outputs conform strictly to defined schemas, eliminating unpredictable text responses. The framework uses 'Agents' as the primary interface, supporting model-agnostic integration (OpenAI, Anthropic, Gemini, etc.) and managing complex components like function tools and dependency injection. This structure ensures reliable, maintainable, and scalable AI workflows for production environments.

- Official technology site: https://ai.pydantic.dev/
- Public AI Tinkerers demos and talks: 8
- Result page: 1 of 1

## Recent Public Talks and Demos

### [Building a production GTM agent: what broke and how we fixed it](https://boston.aitinkerers.org/talks/rsvp_KyMpqyZSw7M)

We built a GTM agent that runs outbound end-to-end. The agent finds prospects in your ICP, comments on their LinkedIn posts, sends connection requests and messages, and books meetings for you. I can share how the agent works and walk through the architecture, including things like: - how we decide which parts of the workflow should be handled by an agent versus deterministic code - how we store structured state in Postgres so the agent can keep track of prior actions and workflow progress - how we built human-in-the-loop review so users can approve actions before the agent proceeds - how the workflow pauses and resumes across long-running tasks I can also show a few examples of where the system initially broke and what we changed to make it more reliable in production.

- Event context: AI Tinkerers Boston: GTM Agentic AI Launch — 2026-06-29 — Boston
- Public talk page: https://boston.aitinkerers.org/talks/rsvp_KyMpqyZSw7M

### [Wrapping MCPs to beat context pollution](https://prague.aitinkerers.org/talks/rsvp_bShvtr6rBCs)

Working production system that uses clever caching to wrap MCP servers into subagents, reducing token counts by 95+% in tool descriptions and another 95+% in task output tokens that stay in the context window

- Event context: AI Tinkerers Prague June Meetup — 2026-06-12 — Prague
- Public talk page: https://prague.aitinkerers.org/talks/rsvp_bShvtr6rBCs

### [Patent Mining for Engineers: Building an Agentic RAG System for Inventive Problem Solving using TRIZ & AI](https://poland.aitinkerers.org/talks/rsvp_oYsjqZvaY7E)

A pipeline that parses patent PDFs, extracts Technical Contradictions, classifies solutions into TRIZ Inventive Principles, and indexes everything into a vector database. This collection is then feeding an AI Agent that helps engineers solve real inventive problems. The demo starts with a raw patent PDF, submits it live to the processing endpoint, and walks through what gets extracted and indexed. Then, given a real mechanical engineering problem, the agent frames it as a TRIZ contradiction, retrieves relevant patents, and proposes concrete solution ideas, powered by domain knowledge, not just plain LLM generation.

- Event context: AI Tinkerers Poland #3 - Meetup in Wrocław — 2026-05-06 — Poland
- Public talk page: https://poland.aitinkerers.org/talks/rsvp_oYsjqZvaY7E

### [AgentMon: Building Step-Level Tracing + Expert Review for PydanticAI Agents](https://cologne.aitinkerers.org/talks/rsvp_ogg5YcS8QGw)

AgentMon is a lightweight framework I built to make agent behavior inspectable after a run. I’ll demo how a PydanticAI “deep research” agent is instrumented with a simple decorator to capture a structured step trace (tool calls, LLM calls, state snapshots, final output) into Postgres. Then I’ll walk through the post-run augmentation pipeline that maps raw steps into an agent-specific ontology and summarizes long outputs so runs are easier to review. Finally, I’ll show the web UI timeline where reviewers can score/label individual steps and whole runs, and how those annotations feed basic analytics across runs. Early stage but iterating fast.

- Event context: AI Tinkerers Cologne #2: Let's Build. — 2026-01-21 — Cologne
- Public talk page: https://cologne.aitinkerers.org/talks/rsvp_ogg5YcS8QGw

### [Arbiter -- LLM Evaluations with automatic interaction tracking, multiple evaluators, and extensible architecture](https://sf.aitinkerers.org/talks/rsvp_mb7z70EdD84)

I'll demo Arbiter's automatic observability for AI agents using two examples that show what makes it different: you get complete visibility into LLM calls and costs with zero instrumentation. Live demo (5-10 min): Example 1: Interaction Tracking (interaction_tracking_example.py) - Run a simple evaluation (semantic similarity check) - Show the automatic interaction log that appears - every LLM call captured with: - Exact prompts and responses - Token usage (input/output) - Latency per call - Model used - Timestamp - The key point: You didn't instrument anything. This just appears automatically. Example 2: Cost Tracking (cost_tracking_example.py) - Same evaluation, now show automatic cost calculation - Breakdown by evaluator, by model, by token type - Uses llm-prices dataset (100+ models, kept up-to-date) - Show how this scales: batch evaluation cost analysis across multiple outputs - The key point: Know exactly what each evaluation costs before you scale to thousands of agent interactions Why this matters for agent builders: When you're running 10K agent interactions daily, you need to know what's happening and what it costs. Arbiter gives you both automatically. No logging infrastructure to build, no cost tracking to implement - it's just there.

- Event context: AI Tinkerers x Google Cloud SF Meetup — 2025-11-21 — San Francisco
- Public talk page: https://sf.aitinkerers.org/talks/rsvp_mb7z70EdD84

### [AI2's Asta is Accelerating Research](https://seattle.aitinkerers.org/talks/rsvp_JHzEM_VI0oA)

Using AI2's Asta Agents/MCP to streamline scientific literature extraction, enhancing throughput for laboratory applications. I'll briefly discuss the project I'm working on as a student researcher at Seattle Children's, an agentic AI genomics target selection pipeline that finds promising drug targets proposed in literature and flags them for study. It uses Asta agents to perform the iterative literature search, as finding promising papers necessitates self-reflection, changing queries, and dynamic analysis, which would be difficult without Asta agents. I will discuss how I implemented this workflow, show how it runs, and discuss its performance with some open-source LLMs. I'll also mention how implementing this same workflow from scratch in June with PydanticAI was a complete pain and Asta simplified it by leaps and bounds. Main message: It's great how a student researcher like myself can have access to these amazing tools and have an actual impact. LLMs will only become more accessible from here, which is cool.

- Event context: Frontier Builds Demo Night: Experiments at the Edge of AI — 2025-11-13 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_JHzEM_VI0oA

### [Pydantic with MCP](https://sf.aitinkerers.org/talks/rsvp_nbaXvVUdGlQ)

demo by Samuel Colvin of Pydantic

- Event context: Protocol Demo Night at WorkOS (MCP, A2A, AG-UI) — 2025-10-10 — San Francisco
- Public talk page: https://sf.aitinkerers.org/talks/rsvp_nbaXvVUdGlQ

### [Debate Concierge](https://prague.aitinkerers.org/talks/rsvp_4Gu3zYErw-s)

The Debate Concierge is an "agentic" (ugh, I know) process written in PydanticAI that takes a topic and a side, and streams back a full research guide (with parallel async search) + a complete argumentative debate case, with possible zoom-in on each argument. This is a customized version of what all the Deep Research products do, one which is attainable to everyone.

- Event context: AI Tinkerers Prague - February 2025 Meetup — 2025-02-25 — Prague
- Public talk page: https://prague.aitinkerers.org/talks/rsvp_4Gu3zYErw-s

## Related Technologies

- [FastAPI](https://aitinkerers.org/technologies/fastapi) ([Markdown](https://aitinkerers.org/technologies/fastapi.md)) — 181 public demos
- [MCP](https://aitinkerers.org/technologies/mcp) ([Markdown](https://aitinkerers.org/technologies/mcp.md)) — 129 public demos
- [Python](https://aitinkerers.org/technologies/python) ([Markdown](https://aitinkerers.org/technologies/python.md)) — 662 public demos
- [AG-UI](https://aitinkerers.org/technologies/ag-ui) ([Markdown](https://aitinkerers.org/technologies/ag-ui.md)) — 4 public demos
- [Amazon Bedrock](https://aitinkerers.org/technologies/amazon-bedrock) ([Markdown](https://aitinkerers.org/technologies/amazon-bedrock.md)) — 20 public demos
- [Asta Agents/MCP](https://aitinkerers.org/technologies/asta-agents-mcp) ([Markdown](https://aitinkerers.org/technologies/asta-agents-mcp.md)) — 1 public demo
- [Braintrust](https://aitinkerers.org/technologies/braintrust) ([Markdown](https://aitinkerers.org/technologies/braintrust.md)) — 2 public demos
- [Claude](https://aitinkerers.org/technologies/claude) ([Markdown](https://aitinkerers.org/technologies/claude.md)) — 173 public demos
- [Exa API](https://aitinkerers.org/technologies/exa-api) ([Markdown](https://aitinkerers.org/technologies/exa-api.md)) — 1 public demo
- [FAISS](https://aitinkerers.org/technologies/faiss) ([Markdown](https://aitinkerers.org/technologies/faiss.md)) — 17 public demos
- [Gemini](https://aitinkerers.org/technologies/gemini) ([Markdown](https://aitinkerers.org/technologies/gemini.md)) — 188 public demos
- [Haiku](https://aitinkerers.org/technologies/haiku) ([Markdown](https://aitinkerers.org/technologies/haiku.md)) — 7 public demos
- [HTTPX](https://aitinkerers.org/technologies/httpx) ([Markdown](https://aitinkerers.org/technologies/httpx.md)) — 3 public demos
- [LangGraph](https://aitinkerers.org/technologies/langgraph) ([Markdown](https://aitinkerers.org/technologies/langgraph.md)) — 67 public demos
- [LLM](https://aitinkerers.org/technologies/llm) ([Markdown](https://aitinkerers.org/technologies/llm.md)) — 123 public demos
- [Logfire](https://aitinkerers.org/technologies/logfire) ([Markdown](https://aitinkerers.org/technologies/logfire.md)) — 2 public demos
- [OpenAI SDK](https://aitinkerers.org/technologies/openai-sdk) ([Markdown](https://aitinkerers.org/technologies/openai-sdk.md)) — 5 public demos
- [Postgres](https://aitinkerers.org/technologies/postgres) ([Markdown](https://aitinkerers.org/technologies/postgres.md)) — 12 public demos