Keep the tools. Remove the handoffs.
Fleet selects work from your tracker, starts an agent beside the repository, runs your checks, opens the merge request, and waits for CI. Every step is recorded. Only an actual judgment call comes back to you.
Prelaunch · for software teams using coding agentsYour coding agents,
Your coding agents,
working as a team.
Fleet connects your issue tracker, coding agents, repositories, and CI in one supervised workflow. Keep your tools. Give every agent the same path from issue to reviewed change.
LinearENG-482 selected
Slackdecision alertsFleet
prioritizeimplementcheckreviewland
Claude Coderunner · working
GitLabMR !4312 · CI green!One decision needs youMerge request !4312 passed checks and is ready to land.
Agent active beside your codeYour code and model keys stay on your runner
Built around the development tools you already use
GitLab
GitHubLinearSlackClaude Code
Codexplanned
Coderplanned
NotionplannedWhat Fleet does
One workflow across the entire delivery loop.
Coding tools are good at writing code. Fleet handles the coordination around them: what to work on, where it runs, which checks must pass, and when a human decision is required.
Stop acting as the webhook.
When CI fails, the member resumes, reads the job result, makes the correction, and pushes again. When CI is merely slow, it suspends at zero compute instead of holding an agent session open.
Make the next run better than the last.
Review feedback stays attached to the work. Fleet can retain a team ruling, apply it in later runs, and propose a deterministic check when the rule is ready to become a gate.
Product walkthrough
See a real workflow, not an AI promise.
Follow one issue from the tracker through implementation, CI, review, and a final human decision.
Member / velaFL-0192 · active
VvelaPayments backend · day 18working
Linear · ENG-482
Prevent duplicate capture during retry
Selected from the Payments backlog. The issue anchor and acceptance criteria have been added to the member context.
PriorityHighProjectBilling reliability
Issue selectedImplementedCI passedLand decision
A clear boundary
Fleet coordinates the work. Your runner executes it.
The Fleet runner lives beside your repository and model credentials. It dials out to the control plane, starts agent runs, performs local checks, and reports coordination metadata. Source code, prompts, transcripts, worktrees, and model keys stay in your execution plane.
Read the security model →Fleet control planeworkflow state · decisions · integration health
runner polls outbound ↓Your execution planesource code · agents · model keys · local checks
Private previewKeep your coding tools.
Keep your coding tools.
Give them one engineering system.
Fleet is prelaunch and not accepting accounts yet. Join the private-preview waitlist or explore the product and security model.