AI capital allocation
Know what your AI spend is actually buying.
Arbitir connects every AI dollar to the person, team, and workflow that spent it, then puts it next to the revenue it produced. So you can move the budget toward whatever is working.
A complete cost ledger
Every dollar across Anthropic and OpenAI, API keys and enterprise seats, in one place. Not an invoice total, a ledger you can take apart.
Attribution to real people
Spend resolved to a person, a team, and a surface. What cannot be resolved is shown as unattributed rather than quietly dropped.
Spend against output
AI cost per closed-won dollar, per opportunity, per meeting booked. The ratio your board is about to start asking for.
How it works
Four steps, and the first three are set up once.
- 01
Connect
Paste a key or complete OAuth. Every credential is validated against the live API before it is stored, then kept in a vault.
- 02
Attribute
Keys, seats, and user records resolve to people by email. Anything that does not resolve goes into a queue you can clear in minutes.
- 03
Compare
Spend lands next to the revenue metrics your team already reports on, by role, with no new reporting work.
- 04
Act
Set budgets by team or person. Hear about a breach the moment it happens, and keep a record of what you decided to do.
We read usage metadata.
Never your prompts.
Arbitir ingests token counts, model names, timestamps, and dollar amounts. It never reads prompt content, completions, or conversation text. There is no code path that could, and that is enforced in the schema rather than promised in a policy.
Credentials you connect are validated once against the live API and then written to a vault. The application database stores the name of the secret, never the secret. Every connection, every credential read, and every view of a person-level page is written to an audit log.