Insights
Notes from the studio, on applied AI.
Long-form writing, field notes and case studies from our work with clients. Our first pieces are on the way.
- Essay
Why agentic AI needs boring infrastructure
The interesting behaviour of AI agents rests on very unglamorous foundations — queues, evaluation, retries and observability.
Coming soon - Case study
Building a domain-tuned research copilot
How we designed an agent that reads, synthesises and cites — while staying honest about what it doesn't know.
Coming soon - Field notes
Evaluation is the product
Notes on running continuous evaluation programmes for LLM-powered features in production.
Coming soon