Latest from our blog
Fresh posts pulled straight from our publication — updated automatically.
The 5% problem: why most AI never ships
Most AI projects die between the prototype and production. Here is where they get stuck — and how to plan around it.
Evaluation is the moat
If you cannot measure a model, you cannot improve it. A practical approach to building evaluation into the workflow.
RAG in production: what actually matters
Retrieval is the easy part. Chunking, ranking, and grounding are where quality is won or lost.
A pragmatic MLOps starter stack
You do not need a platform team to be reproducible. A lean stack that scales from first model to fleet.
How to scope an AI project
Scope for learning, not certainty. How we structure first engagements to de-risk the unknowns fast.
On-device vs cloud inference
Latency, privacy, and cost pull in different directions. A framework for deciding where inference should live.
Get the occasional deep dive
No noise. Just the engineering and strategy notes we'd want to read ourselves.