Ideas

Forward-looking thinking that emerged from FDE engagement work. Cross-engagement, not client-specific. None of this is a commit-to-build; each entry is a record of an architecture or product direction we've considered, why it might matter, and what it would take.

Filter: 5x or skip

Productizable ideas usually die because they're 1.2x improvements that need new commercial structure to ship. Anything that lands here has to clear a 5x improvement threshold against the status quo. If it doesn't, it goes in Lessons or stays in conversation.

Current ideas

• Voice Agent Testing Primitive, instant CRM sandbox from OpenAPI + path-tracing + traffic capture. Enables voice-agent variant testing without provisioning real test tenants. 5-50x against the alternative of getting real CRM sandboxes.
• Canonical CRM + Declarative Overlay, one canonical CRM model internally + declarative per-CRM overlay DSL. In-call logic targets one shape; capability disparities (e.g., AutoOps SMS cancels) become enumerable and testable instead of scattered. Exploration, not commit-to-build.

How entries get here

1. An FDE engagement produces a piece of infrastructure that solved a specific client problem.
2. The pattern shows up in conversation: "this could generalize."
3. We apply the 5x filter: against what status quo? What's the leverage?
4. If it clears, an entry lands here capturing the architecture and the strategic options.
5. If a real productization conversation opens later, the entry becomes the starting point for a planning track.