Private Draft

The 29 personas behind AI

We’ve organized every stage and persona in the AI supply chain, informed by real recruiting at frontier companies. Click any row to see matching profiles from our talent graph.

Shaped by Industry Experts

← ATOMS & ENERGYUSERS & MARKETS →

← Back

Product Management

Defines what to build and why

Product Management

Summary

Known as: Product Manager, Technical Product Manager, Product Lead

Defines what to build and sequences delivery where model capabilities, not feature specs, drive the roadmap. The AI variant is distinct: non-deterministic outputs require probabilistic quality bars instead of pass/fail acceptance criteria, model improvements land on their own timeline (not sprint cycles), and product decisions require eval literacy — reading benchmark results, understanding capability/cost/latency tradeoffs, and knowing when a model upgrade changes what's buildable. Spans a wide technical range: at frontier labs, PMs sit with researchers and sequence roadmap against training milestones; at application companies, they ship features on top of model APIs and own the quality-improvement loop directly. At platform companies, also owns the developer experience layer: API design, documentation, rate limits, and the packaging that turns model capabilities into products developers adopt.

Specializations

Model-Layer — Owns roadmap at the model layer: sequences work against training milestones, reads eval results and capability benchmarks, and makes prioritization calls that require understanding model architecture tradeoffs. Sits with researchers and engineers; fluent in training progress, scaling behavior, and the gap between benchmark performance and user-facing quality. Concentrated at frontier labs and model-is-the-product companies (OpenAI, Anthropic, Cohere).

Application-Layer — Owns AI-powered feature roadmap at companies building on top of model APIs rather than training from scratch. Workflow design, UX for non-deterministic outputs, model selection (cost/capability/latency), and rollout strategy. Owns the quality-improvement loop directly — defining rubrics, triaging failures from production eval signals, and prioritizing what to fix next — since dedicated eval or measurement teams rarely exist at this layer. Technical depth is in integration and product craft, not model internals.

Platform-Layer — Owns ML infrastructure, eval platforms, and data systems. Internal customers are researchers and engineers.

Where the Work Lives

[1]Substrate

[2]Compute

[3]Intelligence

[4]Systems

Secondary

Aligns technical execution, launch criteria, and cross-functional delivery.

[5]Distribution

Primary

Defines what AI features to build, for whom, and how to measure success in market.

Candidate Archetypes

Tom Banks

OpenAI

AI feature PM

Owns UX outcomes, iteration cadence, launch strategy, and the definition of ‘good’ for users.

Alesia Trudie

Anthropic

Platform PM

Owns internal ML, eval, and data systems as products — with engineers and researchers as customers.

Xing Anh

Notion

Application AI PM

Ships AI features on top of model APIs — owns quality rubrics, failure triage, and the eval-driven improvement loop without a dedicated measurement team.

Company Scale

Early-Stage

Occasional

Growth

Primary

Enterprise

Primary

Founders own product until Series A; growth+ has dedicated AI PMs.

Featured Roles