Analyst's Corner: Your AI copilot is making you a worse product manager

Pankaj Bisht wrote this piece in July 2026 for Analyst’s Corner on Medium. It challenges a commonly held assumption in the product community: that AI writing and reasoning tools make product managers more effective.

The argument is counterintuitive. Bisht acknowledges that AI copilots are genuinely powerful at removing friction — generating first drafts of PRDs, structuring meeting notes, summarizing discovery calls. The trouble, he argues, is precisely that they are so good at this.

The craft of product management has always been built through struggle. When a PM spent time staring at a blank document, they were not just producing text — they were refining their understanding of the problem, contradicting themselves, discovering what they did not yet know. Bisht calls this the “craft in the friction”: the uncomfortable cognitive work that forced clarity and conviction before anything got written.

AI copilots bypass this process. A PM can generate a polished, coherent PRD in twenty minutes without ever confronting the moment where they realize they cannot explain why the product should go in a particular direction. The document looks complete. The thinking behind it may not be.

The practical risk Bisht identifies is subtle but significant. A PM presenting an AI-generated proposal they have not actually internalized cannot defend it under pressure, cannot spot the gaps engineers will ask about, and cannot course-correct when circumstances change. The polish of the output masks the absence of genuine understanding.

This concern is especially acute at the point where a PM realizes, mid-presentation, that they do not believe their own proposal. With AI-assisted work, that realization arrives later — sometimes after the team has already committed to the direction.

Bisht is not arguing against AI tools. He is arguing against using them as a replacement for the thinking process rather than a complement to it. The distinction matters most in early-stage work, where most of the value a PM adds is in working through ambiguity rather than in producing documentation.

The article is useful for PMs who feel increasingly productive with AI assistance but sense something is being lost in the process. It gives language to that unease and suggests that the right use of AI is to accelerate synthesis after the hard thinking is done — not to substitute for it.