ObserveSpend & Usage

Know what every agent costs and who it serves.

See AI spend and usage by team, agent, user, model, and tool across model calls and sandbox time. Find the agent that tripled its spend, cap it, and show finance exactly where every dollar went.

Finance · month-endAI provider invoices3 providers
AI spend this month$68,412Up 41%, and nobody could say why
Attributed to the dollarBy team, agent, user, model, and tool
Where it wentSeptember
research-agentSales · 4 users$9,840 · 3× August
Support14 agents · 212 users$21,300
Top modelclaude-opus-5-5$31,140
Most-used toolsalesforce.search41,200 calls
Cap set on research-agent · $3,000Calls stop at the limit from now on
What you get
Every dollar attributed

Spend by team, agent, user, model, and tool, reconciled to the provider invoice.

Usage you can act on

See which agents people actually use, how often, and which are growing fastest.

Caps that hold

Limits per agent, team, or key stop calls at the limit, not after the invoice.

Attribution

Every dollar traced to who spent it.

Model spend is attributed to the team, agent, user, and model behind every call, next to the business results each workflow delivers.

Spend by teamSeptember
Support14 agents · 212 users$21,300
Engineering22 agents · 42 users$18,900
Salesresearch-agent is $9,840 of it$12,200
Reconciled to the invoice$68,412 · every dollar attributed
By team

Finance gets the breakdown, not a single line.

Spend by team, agent, user, and model, reconciled to the provider invoice, with the one driving the increase in focus.

Usage and adoption

See which agents people actually use.

Active users, runs, and spend per agent show what’s working, what’s growing, and what costs more than it should.

Usage · 30 days356 active users
support-bot212 users · 18,204 runs$8,410
code-reviewer42 users · 9,880 runs$3,807
research-agent4 users · 6,120 runs$9,840
4 users, $9,840research-agent is worth a look
Usage

Users, runs, and spend for every agent.

Spot the agent with four users and a five-figure bill before it becomes next month’s surprise.

Budgets and caps

Set limits once. They hold on every call.

Caps per agent, team, or key are checked by the gateway before each request. At the limit, calls stop and the owner hears first.

BudgetsPer agent, team, and key
research-agentAgent$2,940 of $3,000
Support teamTeam$21,300 of $25,000
ci-pipelineKey$410 of $1,000
research-agent at 98% of its capCalls stop at $3,000
Budgets

Every cap, what’s used, and what’s left.

Review and adjust limits in one place, and know none of them depend on someone watching a dashboard.

Spend & Usage

See where your AI spend actually goes.

Connect your agents, and we’ll show spend and usage by team, agent, user, and model, then set a cap live.