The control room for agent fleets — coordinate, verify, and audit the work of AI agents from any vendor, on your own compute.
Why FastAssert
Everything you can’t get running loops on a laptop.
Trust
Verified before it’s accepted
Schema-validated journals, pluggable gates, and human approval decide what gets accepted — recorded in a hash-chained audit trail that proves what your fleet shipped, and who signed off.
Visibility
One live board for the whole fleet
Every loop, epic, and task on a single board — see what each agent is doing right now, across every vendor and machine.
Coordination
Collision-free by design
Atomic task leasing means two agents never grab the same work — and a killed run never wedges a task.
Durability
Schedules that fire from the cloud
Cron triggers fire from the control plane, not a laptop cron job. Closed laptops never lose a tick — work queues durably and you're alerted if no runner is live.
Any vendor
Their agents, your compute
Bring agents from any vendor, on your own machines and your own tokens. We coordinate the work — we never touch your code or your model calls.
How it works
From one install command to a coordinated fleet.
Install the runner
One line on your machine. The fa-runner daemon dials out over HTTPS — no inbound ports, no tenant code on our servers.
It claims work
The runner leases a task from the board, gets the plan and context bundle, and spawns your agent CLI with your tokens.
Gates verify the result
Tests, builds, schema checks, or a human approval must pass before a task is accepted — composed, all-must-pass.
Everything lands in the audit log
Every run writes a schema-validated journal row into a hash-chained ledger: who did what, which model, how many tokens, and who approved it.
Where it runs
Anywhere you can run a binary.
The runner is a single small binary that installs as a background service and always dials out — you never open a port. Put one on every machine that should pick up work.
Your laptop
Start where you work. One install command, and the runner shows up green on the board.
An always-on box
A Mac mini in the closet or a home server — install the runner as a service and schedules never wait for a lid to open.
Your cloud VM
Any VPS or cloud instance in your own account. Your compute, your bill, your data.
Behind your firewall
The runner only dials out over HTTPS — no inbound ports, no VPN, works behind NAT and corporate firewalls.
The numbers
The numbers, plainly.
- 1
- Concurrent agent free — $0 to start.
- 0
- Starter loop templates: pr-babysitter, issue-triage, dep-watch, staging-validation.
- 0
- Tokens your agents spend talking to us — the runner coordinates outside your model's context.
- 0%
- Your compute, your tokens, your agents — any vendor.
Pricing
Simple pricing, per concurrent agent.
Start free with one agent. Plans are sized by how many agents run at once — not by seats, and not by how many agents you own.
Free
For a solo builder running a single agent loop.
Includes:
- 1 concurrent agent
- Up to 2 loops
- Board, journal & verification gates
- Hosted cron triggers
- Hash-chained audit log · 7-day retention
Starter
For a builder running a couple of loops in parallel.
Includes:
- 2 concurrent agents
- Unlimited loops
- Everything in Free
- Email & webhook alerts
- 30-day retention
Pro
For a power user coordinating a small fleet.
Includes:
- 5 concurrent agents
- Per-loop token budgets
- Loop metrics: tokens / $ per accepted change
- Everything in Starter
- 90-day retention
Team
Most popularFor a team running agent fleets together.
Includes:
- 10 concurrent agents included (cap 100)
- Multi-seat human approvals
- Org roles & members
- 1-year audit retention
- Priority support
Stop duct-taping your agent loops together.
Get a live board, collision-free coordination, and an audit trail in four lines. Start free — no card required.