Cost and capacity become part of engineering decisions
The factory records run cost, capacity and delay so work can be routed to suitable models and stopped when value is weak.
Depends on: E2-M1, E3-M1
Epic E4
Economic and self-improving factory
The factory can reason about cost, capacity, quality and its own improvement opportunities.
Purpose
Engineering choices include the price of model calls, verification effort, delay and maintenance burden.
Current status
Proposed. The brief makes cost and resource allocation central, but this website does not claim the factory optimises itself today.
Milestones
Evidence
Cultivate operator brief
Full 20-section assignment preserved locally with SHA256 e8b55ed458e3bc6347735cccee872f59a0bf863b4d1dddb56ba2400b21b2461f.
Local Gas City runtime and commands
The city was running under supervisor control with Mayor active. Rigs, formulas, packs, queues and GitHub credentials were inspected with the installed CLI.
Risks and open questions
Risks
- Optimising for cheap tokens or issue throughput can reduce product quality.
- Self-improvement loops can reward metrics that are easy to game.
Open questions
- Which cost measures are reliable enough to drive routing decisions?
- Who approves changes to the factory itself?
Linked Wardley components
- Agent orchestration
- Model APIs
- Coding models
- Resource accounting
Next steps
- Add cost and run-quality records once a product experiment exists.