AI is splitting product development roles into 5 archetypes. Which fits you best?

The Five Archetypes from Anthropic

Boris Cherny from Anthropic's Claude Code team observed that traditional roles like software engineer and data scientist are losing meaning, and proposed five new archetypes: prototypers, builders, sweepers, growers, and maintainers. The prototyper role, where vibe coders spend most of their time, is becoming a crowded market because anyone with design sense and a Claude Code subscription can now go from idea to working prototype.

Why This Map Is 20 Years Old

Jo reveals that this insight has been reinvented three times over the last 20 years, meaning we already know how this plays out. The first map comes from Simon Wardley, who created the Wardley Map with a horizontal axis of evolution from genesis to commodity. Wardley coined three archetypes: explorers working in messy genesis territory, villagers growing products after finding product-market fit, and town planners keeping mature systems running.

Mapping Archetypes Across Frameworks

When you squint at both frameworks, prototypers map cleanly to explorers, builders and growers map to villagers, and maintainers map to town planners. The key insight from Wardley is that products move through stages and require different team compositions over time. You cannot keep vibe coders creating prototypes once you find an idea that sticks; you need villagers to grow it, and eventually town planners to maintain it.

Kent Beck's 3X Model and Coding Genies

Kent Beck's 3X model from his Facebook days offers a third perspective with explore, expand, and extract phases. Beck calls coding agents 'genies' because they grant wishes in a perverse way. His key insight: coding genies are very good at the explore and prototype phase, giving the most bang for your buck, but they require significantly more guardrails, handholding, and expertise in the expand and extract phases.

The Junior Engineer Value Flip

Beck offers a surprising insight about junior engineers. The traditional value curve showed a productivity dip during onboarding, then steady improvement until the engineer leaves. With AI assistants, juniors can learn faster by using coding genies as personal tutors to ask questions about products and programming concepts. Beck's conclusion: hiring junior engineers pays off even more than before.

The Vibe Coding Wall

Jo addresses vibe coders directly. The first weekend of prototyping feels magical, but then real users show up, feature 40 breaks previous features, and you hit a wall. The advice for software engineers is clear: do not make expert prototyper your long-term career goal. Instead, invest in villager-type skills like system design, architecture, and loop engineering that span the boundaries between archetypes.

The Meta Evolution of Sweeping

The sweeper archetype is evolving rather than disappearing. Sweepers will not manually refactor code anymore; instead, they will build the systems, PR review skills, test harnesses, and guardrails that do the sweeping automatically. Jo is already teaching his agents to become better at sweeping by writing PR review skills and installing test harnesses. The job moves from sweeping the broom to building the sweeping system.

Two Questions to Find Your Place

Jo closes with two questions for viewers. First, when you are at your best and in the zone, which archetype are you actually operating in? Second, looking at your product's current stage, is that archetype what your product actually needs right now? If you are a prototyper in an extract context, that mismatch might require developing new skills or finding a different role.