Ship faster.
Stop redoing the same work.
Fleet solves three problems: autonomous delivery, review and sign-off with a full audit trail, and multi-agent coordination.
Three outcomes that matter
Fleet is organized around outcomes — ship the work, check the work, stop carrying the handoffs yourself — not technical features.
Governed Code Delivery
Agents that write, review, and merge code through an explicit saved workflow. From ticket to governed merge with a full run history.
- Multi-stage workflows with human approval gates where you need them
- A review step runs on every PR routed through the saved workflow
- Automated release management with merge gating and label tracking
- Full audit trail on every decision for compliance and post-mortems
Review, Sign-off & Audit Trail
An independent review step checks the output, an approval gate holds it for your sign-off, and every decision lands in a full audit trail you can query later.
- Per-agent run tracking with cumulative time and session counts
- Full audit trail — every agent decision logged for compliance
- 6-dimension eval scoring, plus a separate risk model that auto-quarantines high-risk agents
- Org hierarchy mirrors your structure with role-based access control
Multi-Agent Coordination
Saved workflows make every dependency, retry, approval, and agent handoff visible across repositories.
- Typed workflow steps carry files, tickets, and PRs through the graph
- Label and schedule triggers start explicit saved definitions
- Bounded outcome routes return flagged work to the right step
- Human approval gates pause irreversible publish and merge actions
Turn the signal into the next move.
Map analytics, reliability, delivery, or manual evidence to an outcome. Fleet ranks what needs a decision, explains whether the problem is the result or the evidence, and starts the right saved workflow from that context.
Evidence
Use a numeric integration metric or start with a manual observation.
Decision
Work the ordered queue: evidence problems, needs attention, then healthy.
Intervention
Start the bound workflow with the KPI evidence carried into the run.
From ticket to merge — one governed workflow
A label can start the saved definition; the graph owns every step, retry, and approval.
From ticket to merge with an explicit human gate and a full audit trail at every step.
Works with the tools your team already uses
Fleet doesn't replace your team's AI tools — it adds the management layer your team is missing.
26 tool-sized entry points for Claude Code teams.
The docs list the tools. The marketing pages map each tool to the developer problem it solves: Brain Daemon insights, Fabric coordination, task routing, configuration, and local agent control.
inspect agent quality and risk
Return evaluation, risk, thoughts, and recommendations.
fleet_fabric_publishpublish a Fabric event
Publish a task, handoff, blocker, decision, update, request, or review event.
fleet_task_assignassign and start a task
Set FLEET_TASK_TITLE and start the selected agent immediately.
fleet_agent_startstart an agent
Start an enabled, stopped agent by fuzzy name.
fleet_logquery the audit timeline
Read the unified Fabric and agent activity timeline.
fleet_statuscheck Fleet health
Get agent counts, pending deliveries, and Brain alerts.
The difference is dramatic
See how Fleet transforms everyday engineering workflows.
Without Fleet
- xPR reviews bottleneck your best engineers for hours every day
- xNo one knows which AI agents are running or how long they have run
- xDeployments depend on manual checklists and one person's availability
- xYour team wastes time on triage, reviews, and handoffs instead of building
- xNo visibility into agent run time or activity across teams
- xAgents work in silos with no shared coordination across repos
With Fleet
- A saved reviewer step checks every PR routed through the workflow
- The dashboard shows agent status, run counts, run time, and decisions
- Saved workflows gate publish and merge actions with human approval where you want it
- Engineers write code. Agents handle triage, review, and shipping.
- Per-agent run tracking — see time spent, total runs, and set time limits
- Saved workflow graphs coordinate agent handoffs across repositories
Explore use cases
Concrete jobs teams run Fleet for — each with the agents, config, and guardrails to set it up.
Ready to take control of
your AI agent fleet?
Install one Go worker binary with no Docker stack, connect the hosted dashboard, and run workflow steps on your infrastructure.