Fleet Engineering vs Multi-Agent Frameworks

June 18, 2026 · View on GitHub

Fleet engineering is governance for populations of agents. Frameworks are execution for individual runs or graphs.

QuestionFleet engineeringLangGraph / CrewAI / AutoGen
Unit of designGoverned populationAgent graph or crew
Primary artifactRegistry + manifestCode / graph definition
"Who did this?"Accountability testTrace per run (if configured)
Permissionsclone / run / edit across teamUsually per-deployment
EconomicsPer-agent caps + attributionPer-run metering (varies)
Human gatesShared inbox across agentsPer-node interrupt (varies)

When to use what

SituationStart with
Building one autonomous workflowloop-engineering
3+ agents across Slack, IDE, SaaSFleet engineering (F1 registry)
Enterprise LangSmith deploymentLangSmith Fleet + this repo's checklists
Custom Python agent graphFramework for execution + fleet manifests for governance

They compose

Framework (execution)  →  Loop (cadence)  →  Fleet (governance)

Register each loop-backed agent in agents/manifests/. Link loops: to loop-engineering patterns. Run fleet-audit before scaling L2+ loops.

Anti-pattern

Using a multi-agent framework instead of a registry when the real problem is "we don't know what agents exist." Frameworks do not replace catalog, inbox, or budget discipline.