Taste & Craft: A Conversation with Tuomas Artman, CTO Linear & Gergely Orosz, @pragmaticengineer

Tuomas Artman, Cofounder and CTO of Linear, posits that the pendulum of software development may be swinging too far in the wrong direction with the rise of powerful AI agents. The newfound ability to ship any feature request almost instantaneously creates a significant risk: building convoluted software with a poor user experience. This echoes Steve Jobs’ philosophy that great products emerge from saying “no” to most ideas. Previously, the inherent difficulty and time cost of engineering served as a natural gate, forcing teams to think critically before building. With AI, that gate is gone, making it dangerously easy to say “yes” to everything.

The New Competitive Landscape: Quality Over Speed

This situation creates an analogy to the hyper-growth era at companies like Uber, where the market dynamics of “winner-takes-all” forced an immense pace of shipping to out-compete rivals. While effective for market capture, it often came at the cost of quality. Today, AI gives every team, even a single person, the capability to ship a vast feature set quickly. In this commoditized landscape, the key differentiator and competitive advantage shifts from the quantity of features to the quality of the software.

Over time people will pick the one that is of higher quality… If you build a bad quality product, you open up yourselves to be… slowly overtaken by the competition.

Measuring quality can be difficult, especially when metrics are tied to revenue. At Uber, a feature that lowered prices might increase revenue, making it hard to justify investments in pure quality. However, a lack of quality creates a vulnerability. When competitors achieve feature parity, users will gradually migrate to the product that offers a better, smoother, and more reliable experience. This erosion of the user base is a slow, gradual slip that won’t be caught by any single A/B test.

Cultivating a Culture of Quality at Linear

Linear has implemented specific, deeply ingrained practices to ensure quality is not an afterthought but a core part of the engineering process.

The Zero-Bug Policy

This policy is a cornerstone of Linear’s engineering culture. It operates on a simple but powerful principle:

• When a bug is reported, an agent automatically assigns it to the relevant engineer.
• This bug becomes that engineer’s highest priority. They are expected to drop everything else to address it.
• The goal is to fix every bug immediately, often within hours, and typically within seven days.

The logic behind this is that the rate of bug creation in any company is relatively constant. Most companies let bugs accumulate in a backlog, creating a “bug debt.” Eventually, they are forced to dedicate time to fixing them, meaning they are still fixing bugs at the same constant rate, just months later. Linear chose to pay a one-time cost—three weeks of no new feature development—to clear their backlog completely. From that point on, the small, continuous effort of fixing bugs as they appear has kept the product at a near-zero bug state. This has the added benefit of delighting users, who are often surprised to see their reported issue fixed within hours.

Quality Wednesdays

This is a unique weekly ritual at Linear. The entire remote engineering team gathers for a 30-45 minute meeting where every engineer presents a small “quality fix” they worked on.

• It’s not a bug fix. A quality fix is a small, subtle improvement—a one-pixel alignment, a smoother animation, a more efficient backend process.
• Engineers must find the problems themselves. This is a critical aspect of the practice. By being required to find and fix a quality issue each week, engineers are conditioned to constantly be on the lookout for imperfections in the product.

This practice started when Tuomas noticed engineers were missing small but crucial UI details, like the specific hover-in/fade-out animation that makes the app feel “fast” and “smooth.” An initial exercise on a tiny part of the UI revealed 35 small problems the team hadn’t noticed. “Quality Wednesdays” institutionalized this process of discovery. The powerful side effect is that engineers, now trained to spot these details, are far less likely to introduce new quality regressions in their own feature work. To date, Linear has fixed over 3,000 of these small, minute details through this process.

The Limitations of AI in Building Quality Software

While AI is incredibly powerful for accelerating certain tasks, it has significant limitations when it comes to product quality.

• Automating Bug Fixes: AI excels here. At Linear, 10% of all bugs are already fixed automatically by an AI agent that creates and lands a pull request without human intervention. This number is expected to approach 100% in the coming years.
• Lacking “Taste”: AI currently has no sense of taste, aesthetics, or human feeling. It cannot understand what makes a user interface feel intuitive or an animation feel natural. It can execute the steps to create a UI, but it cannot imbue it with the thoughtful design that defines a great user experience.
• No Concept of Time: An AI agent doesn’t “feel” the frustration of a two-second delay. It can be told that one second is better than two, but it lacks the human perspective to understand when a performance issue crosses the line from acceptable to annoying.

The Future is the Product Engineer

To thrive in this new AI-powered era, engineers must evolve. As AI commoditizes the act of writing code, the most valuable skills will shift.

• Everyone becomes a product engineer. Engineers will need to be more than just coders; they will need to be mini-PMs who are deeply engaged with the customer, understand their needs, and can translate those needs into well-designed, high-quality features.
• Hiring for Product Sense: Linear’s hiring process is designed to find these individuals. It includes a paid, week-long project where candidates build a feature from scratch, allowing the team to assess their product sense and drive.
• Customer Immersion: Once hired, engineers are immersed in customer feedback through shared Slack channels with major customers and access to a searchable library of all recorded customer meetings. This direct exposure ensures the entire team feels the customer’s pain and understands their needs firsthand.

For engineers looking to grow this “product sense,” the advice is to get closer to customers, build and ship your own projects to gain firsthand experience, and study foundational design principles, such as Apple’s Human Interface Guidelines.

• Tags:
• AI Agents
• Product Development
• Engineering Culture
• Software Quality
• Linear