Know the cost before the work runs.
Set a cost limit per task, per agent and per agent type. Every model call is checked against it before it is made. AI spend becomes a line you can forecast and cap.
The problem
The same task cost more this quarter. Nobody on your side changed a thing.
The model changed, the way it works changed, or the way the vendor counts what you owe changed. All three are theirs. Most companies find out from the invoice, a month later, and cannot report cost per decision, only one blended bill.
What you get
Budget a decision, not just a licence.
Checking a refund is worth a few cents of compute. Reviewing a master agreement is worth a great deal more. Set the limit to match the work.
Estimated before it runs
The cost of a task is estimated before work starts. If it would go over the limit, it stops and asks the owner.
Limits at three levels
Per task, per agent and per agent type. New agents of a type inherit its limit.
Checked before every call
Each model call is checked against the limits before it is made, so spend stays inside what you set.
Cost on the record
Tokens and cost for every call, by agent, model and run, next to the decision it paid for.
Your choice of model
Pick a model per agent. Switching is a config change, not a rebuild. New models from your providers appear without an upgrade.
Failures stop instead of looping
A step that keeps failing is stopped and handed to a person. No retries all night.
Why your CFO will care
Move the meter to your side.
- A line you can forecastLimits per task and per agent turn AI spend into a number you plan, not a surprise.
- Cost per decisionEvery decision carries what it cost, so you can see what each process is worth.
- No markupLeafCraft sells no models and no compute. You use your own model accounts and keys.
Across every agency
Spend, sessions and errors at a glance.
The Control page shows LLM spend per agency, and one switch pauses any of them.
Questions teams ask
Who absorbs it if token prices rise?
You see it before it runs. Each task is estimated against its limit first, and you can switch the model for any agent with a config change.
What happens when a limit is reached?
The work stops and asks the owner you named. It never runs on silence.
Do you resell model usage?
No. LeafCraft sells no models and no compute. You connect your own providers and keys.
Can we see cost per agent from another platform?
Yes. If an outside platform doesn't report its costs, LeafCraft estimates them so the agent stays inside its budget.
Discover every agent
You can't govern what you can't see.
LeafCraft connects to your ecosystem and finds every agent you run: the ones vendors shipped inside your software and the ones your teams built. Then each one gets an owner, comes under your rules, or is removed.
Explore the rest of the platform
How it works
From work arriving to the record, step by step
ExploreDiscover every agent
Find every agent you run, then own, govern or retire it
ExploreOne team
AI agents and people in the same roles
ExploreYour rules
Roles, access, checks, authority, escalation
ExploreYour memory
Every decision and both reasons, kept by you
ExploreBring any agent
40+ agent platforms, any model, your stack
ExploreSecurity and deployment
Your cloud, VPC or on-prem
ExploreFind out what one decision really costs.
Bring one process. We'll set limits per task and per agent with you, and show you the cost before the first run.