See, attribute, and control what every AI agent costs.

Your seat answers four FinOps questions: what are we spending, who is it for, how much is exposed — and are we on track against budget?

Bottom line

$685 spent across 8 agents; at $116/day the 30-day projection is $3,480 vs a $3,000 budget ($480 over). 58% rides Anthropic, 54% is served from cache, and 42% of spend ($291) flows through risky or ungoverned agents.

Do nextCut the run-rate →Review $291 at-risk spend →

$685

total AI spend

across 8 agents

$3,480

projected vs $3,000 budget

$480 over · run-rate $116/day

42%

spend at risk

$291 risky / ungoverned

58%

provider concentration

Anthropic · 54% from cache

Are we on track? — forecast vs budget

over budget by ~$480/mo

Monthly budget

$3,000

Projected (30d)

$3,480

Spent so far

$685

Run-rate $116/day (avg of last 3 days × 30). At this pace the $3,000 budget is reached in ~1 week — already projected over.

Provider concentration

58% Anthropic

Anthropic

58%

OpenAIest

42%

58% of spend rides one provider (Anthropic) — a pricing / availability concentration to weigh against single-vendor leverage.

Cache savings

54% from cache reads

Cache reads

54%

54% of token volume is served from cache reads — cost avoided by reuse. Estimated $1,879 avoided vs uncached at current run-rate; protect this as you scale.

Governance-tagged spend

42% risky / ungoverned

governed · $394risky / ungoverned · $291 ($122 unchargeable)

42% of spend flows through agents that are risky or ungoverned — money you can't cleanly charge back to a P&L or control.

Your surfaces

Cost Center$685

Where spend goes — model, agent, provider, user.

Showback by User6 users

Attribute spend & risk to a person.

Fleet Overview8 agents

The whole fleet, read as spend.

Lensed to the finance seat. Derived: run-rate from costs.series · governanceTaggedSpend · weeksToTarget — from app/data/c16-fleet.ts.