AI Coding Agents Are Changing Engineering Teams

Deepak Bapat, CTO and co founder at Tabs, joins The Tech Trek to talk about how his team is using tools like Claude Code and Cursor, where AI is helping, and why systems thinking may matter more than raw coding ability as engineering work shifts. Practical Takeaways • AI coding agents are already producing useful production work, but judgment still matters. • Tool choice may be less important than standardizing the expected output. • Messy repos can make AI generated work harder to trust, so cleanup and patterns matter. • The future engineer may look more like a product engineer with strong systems thinking. • Teams may move from debating features to rapidly building multiple versions and testing what works. Timestamped Highlights 00:37 What Tabs does and why contracts create hard revenue workflow problems for B2B finance teams. 02:16 Deepak compares pre AI engineering work with the current shift toward AI assisted development. 05:09 How the Tabs engineering team uses Claude Code, Cursor, and other coding tools in real work. 08:13 Why inconsistent codebases create more risk when teams add coding agents 14:00 The idea that teams can build the same feature multiple ways in one afternoon. 20:53 Deepak’s view on whether the future team needs separate PMs and engineers, or more product engineers. 23:38 A future where software can become more bespoke to each customer because AI changes the cost model. One Line That Stuck “You can build on three different work trees the same feature in three different ways and see which one you like, and you can do it all in an afternoon.” Practical Moves From The Conversation • Keep humans close to the review process, especially when the last five percent still requires taste and judgment. • Clean up inconsistent code patterns before letting agents operate broadly across the repo. • Hire for adaptability, systems thinking, and problem solving, not just past tool familiarity. • Use AI to explore more product options faster, but do not remove the need to ask whether the feature should exist. Subscribe or follow The Tech Trek for more conversations on how technical teams are building, hiring, and operating as AI changes the work.