Thinking, in public
Field notes on design, engineering and growth — written by the people who ship the work. No thought leadership, just lessons with receipts.
Multimodal interfaces: beyond the chat box
When chat is the wrong interface: designing AI products with the right modality — voice, image, structured output, inline suggestions — and how to choose.
Announcements inside the product, minus the noise
In-product announcements without banner blindness: an announcements centre, targeting by behaviour instead of blast radius, and expiry rules that stay honest.
Teaching what the AI can do: feature discovery as curriculum
AI features fail when users don't know what to ask. Designing discovery surfaces — prompt starters, worked examples, capability demos — that teach, not decorate.
When to build a voice interface (and when a form wins)
A decision framework for voice UI: where speech beats typing, turn-taking and latency design, barge-in patterns, transcription honesty, and voice privacy.
Naming a company: our process from blank page to trademark
One naming engagement, told end to end: the brief that framed it, the 340 candidates, the screening gauntlet, and how 'Meridian' survived to become the name.
A/B testing statistics for people who ship
Reading A/B tests honestly: minimum detectable effects, the peeking problem, novelty effects, segment traps, and the decision log that ends the arguments.
Analytics for AI features: what to measure
Chat volume tells you nothing. The instrumentation stack we use for AI features: acceptance, correction signals, task completion, cost-quality overlays, weekly review.
Testing brand voice on people, not vibes
Vibes aren't evidence. Test brand voice on real people with comprehension checks, cloze deletion and the 'could a competitor say this' filter.
Status updates clients actually read
The one-screen weekly note format that kills the 40-slide deck: shipped, next, blocked, decisions needed — with working demos instead of jargon and metrics in context.
Article indexes people actually browse
Most blog indexes are a reverse-chronological conveyor belt. Here's how we design journal indexes that invite real browsing — lead stories, rhythm, and honest pagination.
Chunking is a design decision: RAG retrieval craft
Chunk sizing, structure-aware splitting, metadata, hybrid retrieval and reranking — the retrieval engineering that decides whether your RAG feature answers or shrugs.
Review your content like a portfolio
A quarterly content benchmarking ritual: sort every article into winners, flatliners and sleepers, then give each piece one of three decisions — prune, refresh or expand.