MENUCLOSE
Product · Ledger
Every AI dollar, attributed to the actor that spent it.
Route AI through Tallin and each request lands in one ledger: attributed to its AI actor (the human, app, or agent that made the call), its owner and team, the provider, model, source, confidence, and coverage state. Give every agent a budget with an automatic cutoff, watch runaway spend freeze before it compounds, and keep the evidence behind every total.
What the ledger captures
Per-actor spend that does not flatten the evidence.
Built for finance, security, and AI owners who need a common operating record: one that names the agent, app, or person behind each dollar instead of leaving you a partial provider dashboard.
- Actor
- The human, app, or agent that made the request.
- Ownership
- Owner, team, and department behind each actor.
- Provider and model
- Which provider and which model served the call.
- Timestamp and volume
- When the request ran, and how many.
- Tokens, cost, confidence
- Tokens, cost, and source confidence on the number.
- Signal and coverage
- Signal source and the coverage state it earns.
- Path
- Gateway, provider, billing, and imported paths.
What changes for you
Cap the runaway agent before the bill, not after the argument.
- Budget with a cutoff
- Every agent gets an owner and a monthly budget with an automatic cutoff on routed calls. Tallin alerts before the threshold and freezes runaway spend after it, so an agent looping overnight stops costing money before anyone files a ticket.
- One normalized ledger
- Every API key, Copilot license, and seat lands in one normalized ledger attributed by actor (human, app, or agent) with cost authority and confidence on every row, so finance and security stop arguing about which dashboard is right.
- Gaps as queue items
- Coverage gaps surface as actionable queue items, not red flags you discover during a customer audit or board review.
Ledger view
Totals with the evidence still attached.
The ledger does not hide where numbers came from. Every row keeps the actor, provider, surface, cost, and coverage state together, so you can trace a spike back to the exact agent, app, or person that caused it.
Per-actor analysis
Read spend the way you answer for it.
- Per-actor, cross-provider view
- Compare spend and request volume by actor (human, app, or agent) across ChatGPT, Claude, Gemini, Copilot, API usage, and imported paths, no hyperscaler estate required.
- Budget caps that enforce
- Set a monthly budget per actor on routed calls. Tallin alerts before the threshold and freezes runaway spend at the gateway after it.
- Explainable, refresh-honest totals
- Open any rollup and see which dollars are observed, inferred, enforced, or not covered, including direct provider calls and tool side effects outside Tallin. Failed scheduled refreshes and stale data surface before old evidence becomes a leadership report.
Supported signals today
Start with the sources Tallin can read now.
Each source contributes a labeled signal that resolves back to an actor. Some are authoritative, some are imported, and some are partial by provider design. Direct provider calls or tool side effects outside Tallin are named as not covered, never silenced.
- Tallin Gateway activity
- Anthropic telemetry
- OpenAI API telemetry
- ChatGPT Enterprise and Edu Workspace Analytics CSVs
- GitHub Copilot usage and seat telemetry
- Microsoft 365 Copilot usage via Microsoft Graph reports
- AWS CUR Bedrock billing with cost-allocation context where available
- Azure OpenAI billing and resource-level Cost Management signals
- Okta SSO System Log discovery for managed AI app access
- Microsoft Entra and Google Workspace sign-in discovery
- Microsoft Defender, Cloudflare Gateway, Netskope, Zscaler, Palo Alto, Cisco Umbrella, Fortinet, Sentinel, and Splunk network or security-log evidence
- Expense, card, network, SSO, workspace analytics, and CSV discovery imports
Put a budget and an owner on every AI actor.
Tell us about your AI actors, provider stack, and the runaway spend or policy gaps you need to close. We will help you decide whether Tallin is a fit for the first cohort.