AI spend you can see, cap and explain.

Priced by consumption, not per seat. Every team, person, key and agent has a budget, and every run shows what it cost.

Book a demo How budgets work
Names and numbers in these examples are illustrative.

AI cost management that follows your org chart.

Budgets go from organization to team to person. API keys, which your own apps use to call AI models, and agents get budgets of their own. Every call is charged to one of them, so every cost has an owner.

An admin or team lead can top up a person's budget for the month.

Organization $3,912of $12,000
  • Engineering $1,420of $3,000
  • Sales team $1,240of $2,000
  • Support team $860of $1,000 alert sent at 80%at 80%
  • Finance team $184of $600
  • Lena R. $12.40of $40 person
  • @finance-⁠analyst $4.96of $50 sponsor: Finance
  • invoice‑sync $31.70of $100 API key

Example

Every run says what it cost, and who paid.

A named agent runs on its sponsor's budget (the person or team that funds it) or on the budget of whoever calls it. A personal agent runs on its person's budget.

@finance-analyst Monday 07:00 run
5 Oct
STARTED BYits schedule · Mon 07:00
nobody pressed a button: its schedule did
SET-UPthe published one
ACCESSGoogle Sheets · read only
AI MODELapproved · your own account
PRIVACY 2 values masked → restored
RESULTSlack · message + sheet
COST$0.31 → Finance team
charged to the Finance team budget
RECORDkept in Agent Activity
the two lines a budget owner reads firstfollow the whole run

Example

Pay for use, not for seats.

A per‑seat license charges for every person, whether they open the assistant or not. With SecondStack you pay for what people and agents use, and model usage stays on your own provider accounts.

Per seat

40 paid · 12 active

Every seat is paid, every month.

By consumption

40 people · paid by use

Light users cost little. Agent runs land on the team that runs them.

Illustration. You get two bills: your AI providers bill model usage on your own accounts, and SecondStack bills by consumption. Pricing on request: ask us for the current terms.What a model costs before you allow it: Model prices 

Questions finance teams ask.

Do we get two bills?
Two, yes: your providers bill model usage on your accounts; SecondStack bills by consumption, not per seat.
Is there a per‑seat fee?
No. Pricing is by consumption, not per seat, and model usage stays on your own provider accounts. A person who rarely uses it costs little; a team that runs agents all week shows up on its own line.
Who pays when an agent runs?
Its sponsor, the person or team that funds it, or whoever called it. A personal agent's runs go on its person's budget. Every run's cost is on its record.
Can we give each team a budget?
Yes. Budgets go from organization to team to person, with budgets for API keys and agents. They take effect the moment they are saved, and alerts go out when spend passes a threshold you set.
How do we see what a model costs before we allow it?
Model prices shows the list price of each model per cloud. In ControlTower, the admin console, a model has to be on a team's allow‑list before anyone on that team can call it, so the price is known before the spend starts.

The same platform, seen from another team

IT & security Where it runs, what leaves and who can do what. Business teams What agents do for your team every week. Platform engineering Agents for the whole company, controls for your team.

Bring last month's AI invoices.

We show how that spend would map onto budgets by team, and which line a single agent run would land on.

Book a demo