Practical perspectives from the engineers, architects and designers building software with our clients every day.
Our discovery framework, updated for the Cursor + Claude + v0 era the workshops we run, what we deliver, and the artefacts that survive.
What auditors actually ask for under SOC 2 and the EU AI Act and how to get compliant without freezing your AI roadmap for a quarter.
Async writing is the single highest-leverage habit in a distributed engineering org. Here’s the template our AI-augmented teams actually use.
How we cut a client’s AWS bill 38% with Karpenter, OpenCost and Spot instances and the operating model that keeps it there.
A practical look at RAG, agent loops, eval harnesses and cost controls for enterprise teams putting Claude, GPT-4o and Llama 3 in front of users.
Five rules of thumb for keeping Kafka, Redpanda and Iceberg-based architectures debuggable as your event volume grows past the toy stage.
