TFD-0006: Request Type Routing Model

TFD-0006: Request Type Routing Model

Date: 2026-03-18 Status: Accepted Decided by: Board of Directors Facilitated by: Oscar (CEO) Consulted: Oscar, Pablo, Camille, Dana

Context

The board asked "how do we test the production line?" with a real STM request: "capture requirements and produce design spec for Azure Dataset." This exposed a fundamental gap — the request was not a factory request at all. It was work for the deployed talent to do, not the factory.

The factory builds digital talents. Deployed digital talents do work for clients. These are different things with different routes.

Decision

Four request types, each with a distinct routing path:

Type Description Route
Type 1 New talent, new client Full 9-stage pipeline
Type 2 New talent, existing client 9-stage pipeline, abbreviated intake (client context already known)
Type 3 Enhancement to deployed talent Lightweight pipeline — delta only (requirements, build, QA, deploy)
Type 4 Work request for deployed talent Goes directly to the talent, NOT the factory

Client Context Registry

To support Type 2 (abbreviated intake) and track deployed talents per client, a client registry is established at production-lines/clients/{client-slug}/:

  • profile.md — Company profile, contacts, systems, methodology, tools
  • deployed-talents.md — Registry of deployed digital talents
  • history.md — Engagement log

Type 4 Clarification

When a client says "capture requirements for Azure Dataset," that is the deployed EA talent doing its job — using its /ea-exigences-intrant skill. The factory is not involved. The factory only activates for Types 1-3.

Consequences

  • Factory pipeline now handles 3 request types (1, 2, 3) with clear routing
  • Type 4 requests are explicitly out of scope for the factory
  • Client context persists across engagements, enabling abbreviated intake for returning clients
  • Enhancement pipeline (Type 3) template deferred until first Type 3 request arrives