Cast is a structured registry of 240+ named thinkers — thought leaders, archetypes, peers — each with real substance behind the name: voice patterns, mental models, stances, tensions. Not a list of labels. A roster you can actually convene, query, and think with.
Most systems that claim to use "expert personas" are faking it. They attach a famous name to a generic output and call it a point of view. Cast is different: every entity in the registry carries a corpus — actual mental models, documented stances, captured vocabulary, recorded tensions. When Voices convenes a critique panel and seats Marty Cagan next to Teresa Torres, both voices render from real material, not from a model's approximation of their reputations. The difference is whether the room is populated or merely labeled.
The registry is taxonomically coded — O*NET roles, NAICS industries, controlled vocabularies — so you can filter for exactly the perspective your problem needs, not whoever happens to be top of mind.
Named humans, composite archetypes, or inner-circle peers — Cast supplies the identity layer. Voices and Forge consume it to run critiques, build personas, and render judgment channels in each voice's actual cadence.
Every voice carries its corpus into the room: decision patterns, tensions, characteristic vocabulary. The rendering is grounded — each thinker pushes back in the way they actually would, not in the way the model assumes they would.
Cast enforces a three-layer contract on every entity it holds. Identity is closed-world: structured, enumerated, taxonomically coded — findable and deduplicated. Corpus is open-world: the richer the better, no vocabulary constraints. Rendering is downstream: Cast declares identity, Forge and Voices project from it. The chain is one-way — renderings are projections, the registry is upstream — so you can re-render any voice as the corpus deepens without touching the consumers:
Who the thinker is — closed-world, enumerated fields, controlled vocabularies. Powers findability, deduplication, and cross-referencing across the whole system.
What the thinker actually thinks — open-world substance. Voice patterns, mental models, stances, tensions. The raw material that makes rendering real rather than performed.
The first diagram traces how a single entity enters Cast — from source harvesting through identity structuring and corpus building to the rendered voice deposited downstream. The second shows where Cast sits in the portfolio: what feeds it, what it powers, and the Intent governance loop that keeps identity and substance coherent.
Cast wasn't populated from a list of names. It was built entity by entity — identity verified, corpus harvested, substance synthesized — governed by the same machinery that built every product in this portfolio. Slip behind the veil to see the operating model that produced it and the system it sits inside.
Cast carries its own .intent/ governance — signals, specs, decisions, traced in git. The identity/corpus/rendering boundary is a ratified spec. The intake pipeline is a six-stage governed process. Cast itself was built under the same operating model it now depends on.
Cast occupies the Identity position in a larger coherence stack — the upstream registry that Forge renders into deployable personas and Voices draws on to seat critique panels. See where it sits among the other implements.
See the overhead view →