Blog
Field notes from building production AI.
What our engineers learn shipping LLMs, agents and data systems into real organisations — written for people who build.
Template page — replace the placeholder posts below with real articles, or connect this list to a CMS / MDX content source.
Engineering2026
Shipping your first agent to production without regret
The guardrails, evals and observability that separate a demo agent from one you can trust with real users.
ReadRAG2026
Retrieval that actually retrieves: beyond naive vector search
Chunking, hybrid search, re-ranking and evaluation — the parts of RAG that decide whether answers are any good.
ReadMLOps2026
Evaluations are the product
Why we build the eval harness before the feature, and how it keeps LLM systems from silently degrading.
Read