Medium: 21 AI agent use cases that make PMs more productive

What the article is about

Aakash Gupta, who writes one of the most widely read PM newsletters and hosts a podcast with over 65,000 listeners, published a systematic survey of where AI agents create the most value for product managers. The article opens with a specific claim — most PMs can recover 15 to 20 hours weekly through agent automation — and then works through 21 use cases organized into four categories: communication and documentation, research and intelligence, data and analytics, and go-to-market work.

Context

Published in January 2026, the article draws on Gupta’s advisory work with product teams at companies including OpenAI, Anthropic, and Google. The framing separates what AI agents can do into functional domains rather than tool categories, which makes the inventory easier to map against actual PM work.

Key method and takeaways

In communication and documentation, the most time-intensive tasks for most PMs — email and Slack drafting, meeting notes, PRD generation, release notes, executive updates, changelog maintenance — all have viable agent solutions that Gupta describes as already deployed at teams he advises. In research and intelligence, competitor monitoring, user feedback synthesis, interview analysis, and market trend identification appear as use cases where the agent handles aggregation and the PM handles judgment. In data and analytics, the article describes agents that generate automated experiment reports, flag metric anomalies, and surface cohort insights without requiring manual SQL or dashboard navigation. In go-to-market work, agents draft sales battle cards, landing page copy, and case study outlines from existing product information.

The core argument is not that agents replace PM judgment, but that they remove the category of work that involves assembling and formatting information that already exists somewhere. The gap the article identifies — most PMs use none of these workflows — is presented as a temporary advantage for early adopters, with the note that the advantage will compress as adoption widens.

Who it is useful for

Product managers who are building their own agent workflows or evaluating which tasks to automate first. The four-category structure maps directly onto how most PM days are organized, which makes it practical to identify where the largest time sink sits and start there. Particularly useful for PMs at growth-stage companies where process patterns are still forming and there is room to design how work gets done before habits solidify.