Skip to content

Insights

Notes from production systems.

No thought leadership. Write-ups on the architecture decisions we actually make, and the ones we have had to undo.

Latest · Cloud & DevOps

The Cost Architecture of an LLM Product: Cache First, Route Second, Optimise Last

Where the money actually goes in a production LLM system, why the usual optimisation order is backwards, and the gateway architecture that makes spend a property you control rather than discover.

· 7 min read