B17 · orchestration · Orchestration & durability

Orchestration Comparison

Status

planned

Weeks

22–24

Permission gate

none

Depends on

B10 · B11 · B12

Page depth

outline

This is an outline, not a result.
Nothing here has been built. It states what the block is for, what it has to produce and what would make it a failure, so that the claim can be checked against the result later. Written 2026-09-01, and it will change.

Why this block exists

Scheduled here because it comes needs real traffic (B11), a measurement layer (B10), and cost instrumentation (B13). Running this on synthetic scenarios would have produced numbers nobody should trust.

What it has to produce

Determine which multi-agent architecture actually wins, measured on real traffic rather than asserted from preference. This is your stated priority at full strength.

01What it has to teach me

topicrouter / manager-with-tools / handoff / deterministic-with-exceptions topologies

topicfan-out economics

topicmarginal value analysis

02Evidence this page will carry

emptyPrimary artifact

emptyMeasured result

emptyRejection record

2026-09-01Outline created from the AI Operations Roadmap (B10-B27).

← B16 Brownfield ReplacementB18 The Deterministic Finding →