Linear CTO warns AI makes it too easy to ship bad software

At the AI Engineer conference, Linear cofounder and CTO Tuomas Artman sat down with Gergely Orosz (@pragmaticengineer) to discuss what they see as a dangerous new trend: AI agents enabling teams to ship software faster than they can think about whether they should. Artman argues that engineering difficulty previously served as a forcing function for deliberate product thinking — because building was slow and costly, teams had to carefully evaluate what was worth building. With tools like Claude Code (mentioning Opus 4.5 specifically), that friction is gone, and the default behavior becomes shipping everything, resulting in convoluted products that no longer serve end customers well. He invokes Steve Jobs' philosophy that great products require saying no to the vast majority of ideas.

He sees AI creating a similar dynamic across the entire industry, where even a small team or solo developer can match a larger company's feature set.

Artman draws a direct analogy to his time at Uber, where hypergrowth and winner-takes-all competitive pressure meant shipping at all costs — an experience he describes as one he never wants to repeat. He sees AI creating a similar dynamic across the entire industry, where even a small team or solo developer can match a larger company's feature set. In that environment, he argues, "tasteful" and high-quality software becomes the true differentiator. The conversation also covers Linear's internal practices — including "Quality Wednesdays" and a zero-bug policy — as well as the future of software engineers evolving into "product engineers" who bring both technical skill and product judgment to their work.

Key facts

1. 01Tuomas Artman is cofounder and CTO of Linear; the conversation is with Gergely Orosz (@pragmaticengineer) at AI Engineer.
2. 02Artman argues AI agents make it too easy to ship every feature request, risking bloated, low-quality software.
3. 03He cites Steve Jobs' principle that great products come from saying no to 999 things and yes to one.
4. 04Engineering difficulty previously acted as a natural gate forcing deliberate product thinking before building.
5. 05Artman draws an analogy to hypergrowth at Uber, describing it as a chaotic 'ship at all costs' experience he never wants to repeat.
6. 06Claude Code (with Opus 4.5 mentioned) is cited as a specific tool that has dramatically accelerated shipping capability.
7. 07The talk covers Linear's internal quality practices including 'Quality Wednesdays,' a zero-bug policy, and the concept of 'product engineers.'