Experience nowBook a demo

For CFOs and finance leaders

What does each AI decision cost, and can I show my auditor who approved it?

Most companies see one blended AI invoice and a log held by the vendor. LeafCraft sets a cost limit before work runs and keeps every decision, with both reasons, on your side.

One teamYour rulesYour memoryYour budget
Book a demoExperience now
Product screen · example dataLeafCraft console

What's changing for you

01

AI is one blended invoice

You are asked to fund it without seeing the cost per decision.

02

The same task cost more this quarter

The model changed, or the vendor's meter did. Neither was your decision.

03

Your auditor now asks about agents

Who approved this payment includes the agent that prepared it.

Two risks the layer removes

The agents you can't see, and the model that gets retired.

Agents you don't know about

In your company, agents arrive through vendor updates and team pilots. LeafCraft finds all of them, gives each an owner, and brings it under your rules or removes it.

How discovery works

A model that gets retired

An agent built straight on one model has to be rebuilt and re-approved when that model is retired or repriced. On LeafCraft you switch the model for that agent in config and keep everything else. Months and a budget, saved.

Agent tied to one model
  • The model is retired or repriced
  • Prompts, tests and sign-offs rebuilt
  • Months of work, a new approval cycle
  • The history stays with the old tool
MonthsA rebuild budget
Agent on LeafCraft
  • Pick another model for that agent
  • Switching is a config change
  • Test it against your own record first
  • Rules, limits, owner and memory stay
HoursNo rebuild
TimeHours to switch, not months to rebuild
CostNo rebuild budget, no new approval cycle
MemoryThe record carries over to the new model
Your choice of model

What keeps you up

1

One blended invoice

Cost per task or per decision is hard to report. You find out the price moved when the invoice arrives.

2

The doer is also the checker

If one agent prepares and approves the same payment, maker and checker are the same. Your auditor will ask about it.

3

Evidence held by the vendor

The decision trail sits in the vendor's stack. It is the first thing your auditor will ask for.

Why it matters now

One layer under every agent and every person.

You want two things from AI: a cost you can forecast and a trail your auditor accepts. The layer sets a cost limit per task, per agent and per agent type before work runs, records what every decision cost, and keeps the approval trail with both reasons in your own database. The meter sits on your side.

How the layer works

What LeafCraft gives you

01

Cost limits before work runs

Set a limit per task, per agent and per agent type. Each model call is checked against it before it is made.

02

Cost call by call

See tokens and cost for every model call, by agent, model and run. AI spend becomes a line you can forecast and charge to a business budget.

03

Maker and checker never the same

No one checks their own work. One agent can check another, or a person can check an agent, and the record shows both.

04

Approvals on the record

Every approval and rejection is kept with who made it, when and the reason given, next to the agents' steps.

What changes for you

Today
With LeafCraft
One blended AI bill
Cost per task, per agent, per process
Cost known from the invoice
Cost estimated before the work runs
Approvals spread across tools and inboxes
One record with both reasons, exportable
Vendor holds the meter
You hold the meter. No markup on models.
Silence is never a yesEstimated before it runsNothing erased after it's written
Product screen · example dataLeafCraft console

Ask any AI agent vendor

Five questions for your next vendor meeting. They are the ones your auditor and your board will ask you. Ask them first.

Why the layer must be neutral
  1. What will a single task cost before it runs, and can I cap it?
  2. Who absorbs it if token prices rise during the contract?
  3. Can I see cost by task, by agent and by business unit, not one blended invoice?
  4. Who approved each payment your agent touched, and was it escalated?
  5. Can I export the full record of tasks, decisions, approvals and escalations for my auditor?

How it helps your people

Where your peers start.

Industries

FMCG and consumer goodsWho approved this distributor claim, and can we show why?ReadConsumer durablesWho checks the agent that approves warranty claims and dealer payouts?ReadPaints and coatingsWhich dealer schemes did the agent approve, and on whose authority?ReadLiquor and beveragesCan we prove who approved every scheme, state by state?ReadBanking, NBFCs and insuranceCan we show the regulator who did what?ReadGlobal capability centresWe run agents from five vendors. Who governs them all?Read

Functions

FinanceCan I show my auditor who approved each payment, and why?ReadSales and tradeWho approves when an agent changes a price or pays a claim?ReadSupply chain and procurementWhat stops an agent from changing a supplier order it should not?Read

Questions teams ask

Can we charge AI cost back to business units?

Yes. Cost is recorded by agent, model and run, and limits are set per task, per agent and per agent type. That gives you a line you can charge to each business budget.

What stops spend from running over?

Each model call is checked against the limit before it is made, so spend stays inside what you set. A step that keeps failing is stopped instead of looping.

Where are the records kept?

In your own Redis or Postgres, for as long as you set. They export in open formats on demand.

Bring your hardest question.

We'll answer it on one of your own processes, with the record to show for it.

Book a demoExperience now
Book a demoExperience now