second-brain

tuomas-artman

3 min read

Tuomas Artman

CTO of Linear; practitioner of taste-first software development; creator of Quality Wednesdays and zero bug policy.

Last updated: 2026-04-24

Overview

Tuomas Artman is the CTO of Linear, the high-quality project management tool known for its exceptional performance and interaction design. Before Linear he was at Uber, where he experienced first-hand what happens when a company prioritizes growth metrics over quality and where that leads over time.

His central conviction: software quality is a competitive moat that doesn’t show up in any A/B test but determines long-run user loyalty. Linear’s entire development culture — Quality Wednesdays, zero bug policy, week-long paid work trials, customer-exposed engineers — is a systematic operationalization of that belief.

He is skeptical of AI as a taste substitute: “AI doesn’t have taste. They simply don’t. They are not human beings.” The last bastion is AI that can feel a UI — perceive time, detect slowness, respond to animation quality the way a human designer does. Not there yet.

Key Ideas

  • Quality doesn’t appear in A/B tests: quality decay is gradual and unmeasurable in isolation. You only notice when a competitor with better quality quietly overtakes you. No experiment will tell you to invest in it.
  • AI enables over-shipping: the biggest AI-era risk is that cheap execution removes the gating mechanism that forced hard prioritization. Steve Jobs’ “say no to 999 things” becomes harder when you can ship everything.
  • Quality Wednesdays: every Wednesday, every engineer must find and fix a quality problem they found themselves — not assigned bugs. ~2,500–3,000 fixes logged at Linear to date. Side effect: engineers permanently in a quality-detection mode, introducing fewer regressions.
  • Zero bug policy: fixing rate is constant regardless of timing — the only variable is your current backlog. Stop shipping features, burn down to zero, then enforce immediately-assigned bugs as highest priority. Users get fix notifications within hours; trust compounds.
  • 10% of bugs auto-resolved by AI agents at Linear (single-shot, no engineer involvement). Artman expects this to approach 100% over a few years.
  • Product engineer convergence: in 4 years, all engineers will be product engineers — understanding customer needs, shaping UX, acting as mini-PMs. The purely technical role is automated away.
  • Week-long paid work trial: Linear’s hiring process ends with a full paid week of green-field work. Self-selects for people who want to drive a product from start to finish. Few misses since adopting it.

Connections

  • design-taste-craft — extends the taste/craft framework with the engineering practitioner view; AI taste gap, Quality Wednesdays, zero bug policy
  • prototype-and-prune — AI-enabled over-shipping as the failure mode of cheap execution without taste
  • agent-evaluation — quality decay is unmeasurable; no AB test exists for it; gradual competitive erosion
  • agentic-system-failure-modes — 10% auto-resolved bugs as early data point for AI code quality at production scale
  • dylan-field — parallel taste/craft framework from the design angle; both identify taste as irreducibly human

Sources

Graph View

Backlinks