Meridian – Agent FinOps

AI agents are becoming a line item nobody can explain. They call models, tools, and APIs on their own, across vendors and teams, and the bill arrives as one number. Finance can't budget it, allocate it, or defend it in an audit.

Meridian is the system of record for AI labor. Our aim is to make every dollar an agent spends as traceable as payroll: metered at the source, attributed to an owner, and booked to the general ledger. It brings the discipline cloud FinOps gave infrastructure to agent spend, built for the CFO's close, not just the engineer's dashboard.

METER

- Direct API connections to OpenAI, Anthropic, AWS Bedrock, and Microsoft Copilot, plus OpenRouter; Azure OpenAI, Google Vertex AI, and Gemini in early access

- Invoice upload with automated parsing for providers without a billing API

- AI gateway ingest from LiteLLM, Kong AI Gateway, and Portkey

- OpenTelemetry support through an open-source Collector exporter (Apache-2.0) that turns GenAI spans into standard usage records

- GPU capacity feeds for private inference via Prometheus remote_write or OpenTelemetry (DCGM), in early access

CONTROL

- Attribution of spend to agents, teams, and cost centers, aligned with OASA, an open agent-spend attribution specification published under CC BY 4.0

- Cost pools, rate cards, and showback for private, self-hosted inference, with every figure owned by your finance administrators

- Per-organization Slack alerts

PROVE

- GL Bridge posts AI cost allocations directly to QuickBooks Online and Dynamics 365 Business Central

- Journal-entry exports for generic CSV; NetSuite, Dynamics 365 Finance, SAP S/4HANA, and Sage Intacct in early access

- A versioned Metrics Export API, with a native LogicMonitor integration

Start with a free two-week Agent Spend Assessment: connect your usage sources and see where your AI spend actually goes before you commit.