How Software Engineering Is Changing

Software engineer Neil Kakkar compares his work before and after joining Tano, an AI-native startup, using records from three jobs. He reports merging about 11 times as many pull requests per month at Tano as at PostHog, where his output was 2.4 times his earlier rate at Bloomberg. The work covered similarly broad product areas, but what took four years at PostHog was compressed into eight months at Tano. Kakkar attributes much of the increase to AI agents that can build features, investigate data, and prepare changes for review, often running for one or two hours while he handles three to five tasks in parallel. He notes that pull-request counts are an imperfect measure, but says they reflect a substantial change in his output.

The shift also depends on engineering culture: AI alone would not deliver the same gains in a workplace where human approvals and review queues remain bottlenecks. At Tano, engineers take end-to-end ownership, make many decisions independently, and rely on automated code-review agents; human review becomes optional once an engineer knows the codebase. This reduces coordination and lets one person build across design, frontend, and backend. The tradeoff is that faster delivery can expose flaws, while new bottlenecks emerge: context switching, teammates losing a shared understanding of work, weaker attention to design, difficulty pacing work, and limits on AI plans. Kakkar’s broader point is that teams must adapt their processes and encourage AI experimentation to benefit from the tools.