Presenters

Source

🚀 Beyond the Feature Factory: Why AI Makes “Taste” the Ultimate Competitive Advantage

In a world where AI can generate code in seconds, the biggest risk to software isn’t moving too slowly—it’s moving too fast in the wrong direction. 📉

In a recent conversation between Tuomas Artman (Co-founder of Linear) and Gergely Orosz (The Pragmatic Engineer), the duo explored how the “shipping frenzy” triggered by AI agents might actually be degrading the quality of modern software. They dive deep into how Linear maintains its legendary quality bar and why the future of engineering belongs to those who think like product owners.

⚠️ The Trap of Immediate Shipping

The pendulum has swung toward a dangerous extreme. Because AI agents can now implement feature requests almost immediately, teams face an overwhelming temptation to say “yes” to everything.

Tuomas argues that this leads to convoluted software and a fractured user experience. He references Steve Jobs’ famous philosophy: Great products come from saying “no” to 999 things to ensure the one thing you ship is perfect. 🎯

Historically, the difficulty of engineering acted as a natural “gate.” Because building took time, teams had to think deeply before starting. Now that AI has lowered the cost of shipping to near zero, that gate is gone. The challenge today isn’t just outpacing the competition; it’s building tasteful software that stands out in a sea of AI-generated mediocrity.

🛠️ The Linear Way: Quality Wednesdays

How does a team keep a high quality bar when the pressure to move fast is constant? At Linear, they use a ritual called Quality Wednesdays. 🛰️

  • The Format: The entire engineering team (around 25 people) joins a 30-minute call.
  • The Requirement: Every single engineer must demonstrate one “quality fix” they implemented that week.
  • The Scope: These aren’t major features. They are minute details—fixing a one-pixel misalignment or ensuring a button has a perfect 150ms fade-out to feel “smooth” rather than just “functional.”

This practice started when Tuomas realized that a tiny menu in the app actually had 35 hidden UI problems. Since then, the team has fixed between 2,500 and 3,000 minute details. This ritual creates a psychological shift: engineers are always on the lookout for regressions because they know they need a fix to show off by Wednesday. 💡

🛡️ The Zero Bug Policy

Most companies treat bugs as a backlog problem—something to be “managed” or “triaged” for later. Linear operates on a Zero Bug Policy.

The Rules of Engagement:

  1. When a bug is reported, an AI agent automatically assigns it to the relevant engineer.
  2. That bug becomes the highest priority. The engineer drops everything else to fix it immediately.
  3. The goal is to resolve bugs within 7 days, though many are fixed within 2 to 3 hours.

Tuomas points out a vital truth: the rate of bug creation is constant. Fixing a bug three months from now takes the same amount of effort as fixing it today, but fixing it today keeps the product pristine and the users happy. Linear even spent three full weeks pausing all new feature development just to bring their bug count to absolute zero. 🛑

🤖 Can AI Have Taste?

While AI is incredible at “piping data” and fixing objective defects—Tuomas notes that 10% of Linear’s bugs are already fixed automatically by AI—it lacks a crucial human element: Taste. 🎭

AI does not feel frustration. It doesn’t know if a two-second delay is “too slow” or if an animation feels “unnatural.” Tuomas highlights an experiment by design engineer Emil Widlund, who compared AI-generated animations to human-refined ones. While the AI did the “right” things technically, the human version felt “natural” and “well-designed” because of subtle easing and timing that AI simply cannot yet perceive.

👨‍💻 The Rise of the Product Engineer

As AI takes over the “hard engineering” tasks like infrastructure scaling and boilerplate code, the role of the software engineer is evolving. 🦾

The Shift: We are moving away from the “pure coder” toward the Product Engineer.

  • Hiring at Linear: They use a one-week paid trial where candidates build a greenfield project from start to finish. This filters for people who care about the why, not just the how.
  • Customer Proximity: Every engineer at Linear is exposed to a “fire hose” of customer feedback via shared Slack channels and recorded meetings. 📡
  • The Goal: Engineers must become “mini-PMs” who understand customer pain points and can design solutions, not just execute tickets.

💬 Q&A Highlights

Gergely: Should engineering teams be slowing down now that we have tools like Claude Code and Opus 4.5? Tuomas: No, we want to move faster, but we move faster by automating the “mindless” tasks like bug fixing (aiming for 100% automation there) so we can spend more time on design and user experience.

Gergely: How do you measure quality if it doesn’t immediately affect revenue? Tuomas: You can’t always AB test quality. At Uber, we focused on revenue and trips, and quality didn’t seem to matter—until a competitor like Lyft shipped a better-feeling app. Over time, users slip away to the higher-quality product. It’s a gradual slip, not a sudden drop. 📉

🚀 Closing Advice for Engineers

If you want to thrive in the AI era, you must develop your product sense.

  1. Get your hands dirty: Build something for yourself, ship it, and listen to your first users.
  2. Study the masters: Tuomas recommends reading Apple’s Human Interface Guidelines as the “gold standard” for understanding great UX. 📖

In a world of infinite code, quality and taste are the only things that can’t be automated. Make them your superpower. ✨

Appendix