Free
Test governed calls.
$0
/ month
Retry → duplicate → 💥
Dedup → once only → ✅
- 3,000 AI requests / month
- Duplicate protection
- 1 governed tool
Stop infinite loops, block unsafe tools, and audit every decision before it turns into real damage.
Test governed calls.
Run a real agent.
Run production safely.
Control the biggest tool surface.
Full backend feature matrix.
| Feature | Free $0 |
Starter $29 |
Pro $149 |
Agency $799 |
|---|---|---|---|---|
| Core | ||||
| AI volume / monthMonthly limit for AI requests routed through OnceOnly. | 3,000 | 100,000 | 1,000,000 | 10,000,000 |
| Automation volume / monthMonthly limit for automation blocks routed through OnceOnly. | 1,000 | 20,000 | 200,000 | 2,000,000 |
| Continues after limitFree blocks new AI and automation requests. Paid plans continue, but OnceOnly stops enforcing the plan limit. | No | Yes | Yes | Yes |
| Max TTLMaximum time a run or idempotency record can stay active. | 1h | 24h | 7d | 30d |
| IdempotencyDeduplicates retries so the same side effect runs once. | Yes | Yes | Yes | Yes |
| AI runSends agent calls through OnceOnly before execution. | Yes | Yes | Yes | Yes |
| Execution leaseKeeps one long-running execution active at a time. | Yes | Yes | Yes | Yes |
| Policy controls | ||||
| Governed toolsHow many tools you can register for governed execution. | 1 | 20 | 100 | 1000 |
| Blocked toolsBlocks selected tools from being called. | Yes | Yes | Yes | Yes |
| Allowed toolsLimits the agent to an approved tool list. | Yes | Yes | Yes | Yes |
| Actions per hour capCaps total agent actions per hour. | No | Yes | Yes | Yes |
| Calls per tool capCaps repeated calls to the same tool. | No | Yes | Yes | Yes |
| Daily spend capBlocks calls after the daily spend limit is reached. | No | Yes | Yes | Yes |
| Tool pricing rulesAssigns tool costs so spend caps can be enforced. | No | No | Yes | Yes |
| Policy templatesPreset policy setups for faster onboarding. | No | No | Yes | Yes |
| Observability | ||||
| Agent logsDecision history for allowed, blocked, and deduped calls. | No | No | Yes | Yes |
| Agent metricsUsage and outcome metrics across runs and tools. | No | No | Yes | Yes |
| Operations | ||||
| Kill switchImmediately disables an agent. | No | No | Yes | Yes |
Short answers to the questions that matter before you choose a plan.
Each governed AI run through OnceOnly counts toward the monthly AI volume in your plan.
Yes. Every plan supports AI runs. The difference is volume, policy depth, observability, and how many governed tools you can register.
Starter adds much higher AI volume, more governed tools, and daily spend caps. Free hard-stops at the plan limit; Starter continues past it.
Choose Pro when the agent is live in production and you need logs, metrics, tool pricing rules, and more governed tools.
Choose Agency when you need the highest limits, allowed tools, kill switch, and control over a much larger tool surface.
OnceOnly blocks the call before execution and records the decision, so the agent does not keep spending or trigger the same side effect again.
No. OnceOnly stores action keys and decision metadata needed for dedupe, policy enforcement, and audit history.