A month-by-month line showing how often work shipped across the current changelog view.

Monthly commit activity

A research brief on the recursive paradox of AI tooling: builders pay labs to dogfood their own product, the next model release eats the middleware, and only proprietary datasets and expert human loops survive.

wrapper graveyard

AI is not a replacement for human skill -- it is a multiplier of it. Research on the jagged frontier, cognitive offloading, and the reverse Dunning-Kruger effect reveals that AI widens the gap between those with domain judgment and those without it.

RIP doomposters

A research brief on how frontier labs compress software markets through rapid releases, two-speed model ladders, tapered usage, dated retirements, and fast value re-internalization.

clock speed

Domain expertise, opinionated systems, and the sub-$1K/month moat in an era of trillion-dollar AI. A research brief on the barbell effect crushing the startup middle class.

startups 2026

The social algorithm is becoming a personal choice. Bluesky custom feeds, Threads Dear Algo, RSS revival, and EU regulation are converging to make feed curation a user-level decision. But the geeks-MOPs-sociopaths lifecycle predicts that only 1-9% of users will ever exercise that choice -- and that small percentage will reshape culture for everyone else.

fork social media

A research brief on monetizing fast-decaying AI features through rapid payback pricing, distribution velocity, and an agent harness core that compounds as models improve.

edge harvesting

A research brief on why software builders keep reinventing the wheel: public creation is a labor-market signal, AI makes greenfield rebuilding cheap, and the durable opportunity is the wheel factory, not the wheel.

wheel factories

A research brief on provider-owned core IP, client usage rights, error-cost-based packaging, and explicit handoff artifacts that preserve speed without lock-in.

open-core contracts

A research brief arguing that most teams need a middle layer between intuition-driven prompt edits and full benchmark infrastructure: structured operator evidence drawn from real traces, failure taxonomies, rubrics, and micro challenge sets.

operator evidence

A research brief on the managerial AI rush: why vibe-coding demos prompt premature layoffs, quality regressions, and upstream spend transfers to labs and hyperscalers.

buildooor — book a 4 hour work session before you rewrite the whole thing