Medium: A day in the life of a product manager using AI
Sam Bobo’s article, published on Medium in December 2025, follows one day in the working life of a product manager who has embedded AI into each phase of product development. The piece is not a tool review or a how-to guide — it is a personal account of where AI adds practical efficiency and where it stops short.
Market research
The day opens with market research, where Microsoft Copilot Analyst generates competitive analysis summaries from structured data inputs. The summaries still require human validation: Bobo is explicit that AI research outputs need a review step before they inform any decision. The cognitive shift here is not dramatic — it is a change from reading primary sources to checking AI-generated synthesis against them — but the time savings are real and the output format is consistent.
Product design
The design phase is where the workflow diverges most from traditional practice. Rather than writing a design brief and waiting for a designer, Bobo describes a process of iterating on interactive prototypes in 30 to 45 minutes using Figma Make. The prototype is exploratory, not final. Problem definition remains manual — Bobo frames this as a human process, rooted in Design Thinking approaches that identify the right problem before any interface gets built. The speed gain is in iteration, not in the thinking that precedes it.
Requirements documentation
Requirements documentation is the most systematized part of the workflow. Copilot with a custom style guide converts verbose product descriptions into structured specification templates. The consistency this produces across a team matters more than any individual time saving: the output format becomes predictable, which reduces review friction and downstream misunderstandings.
Marketing and documentation
The marketing step adds AI-assisted content translation and image generation for video production and written materials. Bobo’s handling of this phase carries the article’s central argument: AI outputs here, as elsewhere in the day, are reviewed, validated, and curated before they become anything durable.
Who this is for
The article is most useful for product managers who are past the basics and want to move beyond ad-hoc prompting toward a workflow that handles specific tasks consistently. It is not a framework or an exhaustive taxonomy — it is a working account of one day, which makes it a useful reference point for comparing your own current practice. PMs who are still running each AI interaction as a one-off will find it a practical starting point for deciding where to build in structure first.