Insights
Notes from the build
Field reports, evals, and opinionated takes from the team.
-
What’s new in the AI world — five shifts to watch in mid-2026
A quarterly digest from inside the build: foundation-model vendors as consulting firms, coding agents as infrastructure, the maturing interop layer, cost no longer the bottleneck, and regulation finally…
-
OpenAI bought tomoro.ai. What that means for AI buyers.
OpenAI's acquisition of tomoro.ai signals that AI enablement is now a recognised discipline — and the independent pool of senior delivery partners just got smaller. Field commentary for…
-
Coding agents went from novelty to default in eighteen months
Field report on what changed for engineering teams since AI coding assistants stopped being experimental — throughput, code review, junior development, and the failure modes worth naming.
-
Where AI agents are actually going next
From assistants to operators, single models to routed fleets, point automation to system-of-record participation — the trajectory we see from inside the build.
-
The off-the-shelf AI agent trap
Packaged AI agents take twenty minutes to install and six months to remove. The five failure modes we see most often, and where custom is the cheaper answer.
-
The Director of AI trap
One person, six jobs. Why the senior-AI-hire-as-first-move pattern fails — and the sequencing that fixes it.
-
The hidden cost of your first AI hire
Why hiring a Head of AI is the slowest, riskiest, and most expensive way to actually get AI-ready — and what to do instead.
-
SEO after AI Overviews
Field notes from rebuilding three sites for the post-generative-search era.
-
Evals are the spec
Why we write evals before the agent — and how the eval set replaces the design document.
-
Shipping agents, not demos
A working note on what separates an LLM toy from something on-call engineers can live with.