Insights
Field-tested guides on AI products, product delivery and engineering, written by the people who build them. Practical enough to use on Monday.
Trust comes from how the product behaves when the model is wrong.
A practical playbook for turning an AI demo into a feature customers rely on: tolerable failure, wrong-answer design, evaluations, guardrails, latency, logging and rollout.
Read the playbookStart here
Turn an idea into a first release without buying a rewrite.
Decide what AI should do in your product, and how people will trust it.
Keep a product cheap to change as the team and surface area grow.
AI products, product delivery, engineering and notes from inside the studio.
A practical guide to AI running costs: the per-task formula, a worked example with labelled made-up prices, the levers that cut spend, and the budgets and reviews that keep it predictable.
ReadA decision guide to on-device and hosted AI: seven factors, the memory maths behind local models, quantisation trade-offs, runtimes, and hybrid routing that asks before data leaves.
ReadAn engineering note on agent transparency: Neel's turn line field by field, one engine-measured rate, labelled gaps, an honest inspector, scoped approvals and failures that say what to do next.
ReadWhat to send a product studio, a one-page brief template, how to compare fixed-scope, time-and-materials and milestone proposals, and what week 1 should produce.
ReadA playbook for inheriting or rescuing an app: the first 10 working days, a risk register, stabilise-before-extend rules, refactor-or-rewrite criteria and a handover checklist.
ReadHow we render a sourced, physically lit solar system on a company website without hurting load speed: real data, real maps, HDR light, strict budgets.
ReadA staged, framework-neutral roadmap from idea to production: retire the biggest risk, ship the smallest valuable release, instrument it, expand screens, harden and operate.
ReadA neutral guide to choosing between native, Flutter, React Native, Kotlin Multiplatform and the web: decision criteria, hidden costs, when Flutter fits, and a scoring worksheet.
ReadFramework-agnostic architecture for products past their first release: boundaries, feature modules, state ownership, testing, observability, release engineering and signs it is failing.
ReadPut the thinking to work
Send us the decision and the constraint. A delivery lead replies within one business day.
