Lucas R. Haines
All work

Rebuilding how a PM team works with agents

When
2025–now
Problem
Specs, prototypes and reviews took days or weeks, and context was lost at every handoff between customer evidence, product intent and delivery.
My role
Came up with, designed and built the workflow, then rolled it out to my team of six and two peer PM teams.
Result
Specs and interactive prototypes now take hours or days instead of days or weeks — and design, engineering and leadership get clearer inputs sooner.

How it works

PMs own the parts that need judgment: the customer problem, UX direction, priorities and decisions. Specialized AI agents handle the defined, repeatable steps — drafting customer-grounded specifications, interactive prototypes, structured reviews, experiment plans and roadmap updates.

Every artifact is standardized and versioned, and automated completeness and schema checks keep related work connected. The result is one consistent path from customer evidence, to specification, to prototype, to review, to experiment and delivery planning.

What stays human

Agents don’t decide what to build. People stay accountable for the problem, the tradeoffs and the call — the workflow just removes the setup and coordination work around those decisions.