Transcript Notes
What this chapter covers.
The biggest Telethryve 2.0 leap is orchestration.
When a job is too broad for one straight run, the user can start an orchestrated flow from chat. A coordinator breaks the goal into bounded tasks, exposes budgets, chooses reusable roles, and keeps a Workspace Board updated as the work moves.
With native Codex multi agent support available, the coordinator can spawn subagents for parallel research, review, and verification, then mirror their evidence back into the board. For write work, Telethryve keeps stronger control.
Bridge workers can use roles like frontend worker, tester, reviewer, researcher, and release checker, and write-capable workers can run inside isolated git worktrees. Known operations can also use deterministic dispatchers such as git status, git diff, browser open, MCP inspect, deploy plan, PR prep, and test run.
This is the self-transforming machine learning to organize its own labor.