The Product Team's Evolution, Stage 3: Encoding the Craft

Part 2 of a series on how non-engineering product roles change as teams climb the AI maturity curve. This post takes Stage 3 and follows four example careers through it: a UX designer, a product owner, a business analyst, and a product manager.

Stage 3 is where the product role changes in a way you can see. At Stage 2 a practitioner used an assistant and kept working the old way. At Stage 3 they do something that feels slow at first. They write down how they actually do the work, encode their standards into agents the rest of the team can reuse, and shift their day from producing artifacts to directing and reviewing them. The model calls this stage Task, and the phrase that captures it is to use and build agents that accelerate your own competency. The key word is your own. Stage 3 agents are personal. They carry one person's judgment, made repeatable.

The most useful way to think about a skill or task agent is as a written job description for a unit of work. Here is how we write a story, run a synthesis, model a requirement, or frame a strategy, and here is the context you need to do it well. Writing that description forces a practitioner to make their craft explicit, which is harder and more valuable than it sounds. Most senior product people carry their standards as instinct. Stage 3 asks them to externalize the instinct so something else can execute it and so the team can finally share one definition of good.

This is the first place the floor of competence visibly rises. Once a standard is encoded, it stops applying to one document and starts applying to everything the agent produces from it. A fuzzy opinion was survivable when it shaped a single artifact. Stamped across every output, it has to be right. The practitioner's reach widens, the bar on their judgment goes up with it, and the work gets bigger without getting easier.

A health gauge travels with this stage. We hold a strong Stage 3 to a high acceptance rate on agent output, in the low nineties as a percentage. Below that band, the agent keeps producing things the practitioner has to rework, which means the standard or the context is weak. Above it, the practitioner is rubber-stamping, and downstream defects are quietly accumulating. The number is a thermometer, not a target to maximize.

The business analyst becomes a requirements architect

The analyst encodes elicitation frameworks, requirement templates, and traceability standards into task agents. Feed a stakeholder transcript in, and the agent produces structured requirements that trace back to stated objectives. The analyst stops authoring every line and starts reviewing the output against intent, hunting for the gap the agent papered over and the requirement it assumed rather than heard. The valuable skill becomes requirements architecture and the recognition of ambiguity and contradiction that a model will smooth right past. The agent is fast at structure. The analyst is the one who knows the structure is missing the thing nobody said out loud.

The product owner becomes a backlog architect

The owner encodes story standards into agents: INVEST criteria, acceptance-criteria patterns, the team's definition of ready and done. A feature concept goes in, a coherent story set comes out, and the owner reviews it for the thing no template can supply, which is whether the story is worth building. The agent drafts well-formed stories all day. It cannot tell you that two of them should be cut and a third reframed around a different user need. That judgment is the work now, and it is exactly where the owner's value concentrates as the drafting falls away.

The UX designer becomes an experience director

The designer encodes research-synthesis methods, accessibility standards, and content and voice guidelines into agents. Research notes go in, and synthesized insights and a first-pass journey map come out. They review against the evidence, checking whether an "insight" is actually supported by what users did or whether the model manufactured a tidy narrative. The skill shifts toward research architecture and design judgment, and the floor holds: someone who cannot tell a real finding from a plausible one cannot run this loop safely.

The product manager becomes a strategy director

The PM encodes PRD structure, prioritization frameworks, and the shape of a good strategy narrative into agents. They spend less time writing the document and more time directing intent and judging what comes back, especially the parts where the model produced confident strategic claims with no grounding. The PM is the evaluation point. Where they have a firm standard, they let the agent run and check the result. Where they do not, they still do the work by hand and watch closely, because that hands-on pass is how the next standard gets discovered.

Directing agents starts to feel like managing a team

By Stage 3 a familiar intuition kicks in across all four roles. Directing a fleet of agents resembles managing people. You set expectations, review output, and decide how much to trust each one. The mechanics line up, and the practitioners who already manage or mentor well tend to make the jump fastest.

The intuition holds with one sharp caveat, the same one engineers hit. Agents do not fail where people fail. A capable colleague's error rate tracks difficulty, and they usually know when they are out of their depth. Agents invert that. A story-writing agent can handle a complex epic and then botch a trivial one with total confidence, or produce a beautifully formatted requirements doc that is subtly, expensively wrong. Confidence comes apart from correctness in a way it rarely does with a competent person. So the Stage 3 reviewer cannot coast on the easy artifacts. Easy is where the surprise lives.

What Stage 3 sets up

Stage 3 produces a real jump in throughput and a real change in each practitioner. The analyst, owner, designer, and PM now spend their days defining intent and judging output, backed by a personal library of agents that encode their own standards. The drafting load drops. The judgment load rises, and so does its value.

The ceiling is that all of it is still personal. Each person's agents reflect their own definition of good, so a team of five product owners still has five definitions, and the reviewing is still a human act performed one artifact at a time. To go faster, the team has to reconcile those personal agents into shared workflows and move the checking out of individual heads and into systems that run automatically. That move is Stage 4, where the role becomes a team event and the practitioner becomes an architect of the workflow itself.

How each role shifts at Stage 3