Applying (Pydantic) AI.

I build enterprise AI agents with Pydantic AI. I write about what works in production and where Applied AI is headed, sharing what I learn through open-source tools and patterns.

· Leads AI operations @ Sidetrade

Harness Engineering

I build the capabilities around AI models that make enterprise agents reliable, trustworthy and production-ready.

Always-On Agents

I explore how open-source, always-on agents can move beyond conversations to proactively work, learn and act autonomously.

Factory Engineering

I believe the future of business is AI-native factories, where humans and autonomous agents work together to continuously deliver outcomes.

  • Nothing published yet.

Pydantic AI Factory

About the project

An open-source software factory built with Pydantic AI, one station at a time.

  1. 00
    What a factory needsMissions, stations, gates, methodologies, artifacts and much more.factory-engineering · Pydantic AI Factory
  2. 01
    From signal to approved specA one-line brief for a to-do list page becomes a design that can be questioned and approved.methodology · Pydantic Capabilities, Elephant Goldfish
  3. 02
    Evaluating Design QualityEvals, baselines and the hidden stakeholder.evals · Pydantic Evals
  4. 03
    Building from a SpecThe build station that takes the design hand off from the design station.harness-engineering · Pydantic AI Harness, Coder Capability, OpenSandbox
  5. 04
    Run checks with DaggerChecks as gates, and rework when they fail.quality-gates · Dagger, e2e
  6. 05
    Evaluating Build and QualityBlind handoff and hidden journeys.evals · Factory Evals
  7. 06
    Dynamic build workflowsDoes Monty & CodeMode optimise the factory's token usage?token-optimisation · Monty, Logfire Agent Sandboxes
  8. 07
    End-to-end evaluationProve how well the factory performs, and identify room for improvement evals · Factory Evals, Code mode benchmark
  9. 09
    The learning loopOutcomes and learning improve the stations for the next intent.learning-loop