Blog
Engineering notes on AI, SaaS platforms and building software that ships.
Written from what we run in production across our own products and our clients'. For the long-form pieces, see Insights.
Retrieval
·
7 min
Retrieval Quality Is a Chunking Problem, Not a Model Problem
Swapping embedding models is the cheapest thing to try and almost never the thing that's broken. Here's what actually moves recall.
Tool Design
·
8 min
Why Your Agent Picks the Wrong Tool, and How to Fix It
Most agent tool-selection failures are interface design failures. Here is how we name, scope and gate tools so the model stops guessing.
Confidence Scoring
·
8 min
Confidence Scoring Makes AI Output Reviewable, Not Trusted
How to attach honest confidence to model output, route the low end to humans, and stop shipping features that quietly fabricate.
AI Engineering
·
8 min
Does That Feature Need an LLM or Just a Query?
A decision rule for separating features that genuinely need a language model from features that need a well-written SQL statement.