Field Notes
Field notes from
shipping AI.
Engineering and ops writing from the team — evals, agents, RAG, cost guardrails, and the patterns we ship into every Pro and Custom build. Published when we have something we’ve actually run in production and not before.
- Frameworks
The eval harness is the deliverable, not the feature
Why we ship the test suite before the agent — and what changes when you do.
Priya Ramanathan/Principal, Engineering
April 8, 2026
8 min read
- Decision Tools
When to use RAG vs. fine-tuning in 2026
A practical decision tree, with the questions we actually ask in scoping calls.
Daniel Reinhardt/Partner, Strategy
March 25, 2026
8 min read
- Field Notes
Cost guardrails for production LLM apps
The eight runaway-bill patterns we've watched in the wild — and how to bound each one.
Marcus Whitfield/Director, Operations
March 11, 2026
9 min read
- Frameworks
Why we sell AI solutions at fixed prices
The economics of productized AI work, and why the dominant pricing models are misaligned.
Amara Okafor/Managing Partner
February 26, 2026
7 min read
- Post-Mortems
The audit-first playbook for AI in operations
Before you build anything, audit what you have. Here's the two-week diagnostic we run.
Marcus Whitfield/Director, Operations
February 12, 2026
8 min read