Ramp becomes AI spend manager
$1.5B/year corporate card neolab
This pushes Ramp past card controls and into being the finance system for AI itself. Instead of only seeing the invoice after the month ends, Ramp can ingest usage from OpenAI, Anthropic, OpenRouter, and other sources, map spend by model, team, user, and project, and show which token costs belong in COGS versus OpEx. That turns AI from a black box engineering bill into something finance can govern like any other operating expense.
-
The product fits Ramp's long running strategy of using unstructured finance data to automate back office decisions. Earlier AI features focused on receipts, invoices, contracts, and vendor terms. Token spend is the next data stream, and it matters because model choice, routing, and usage patterns now move budgets as much as headcount or software seats do.
-
What makes this useful is not just cost tracking, but normalization. Ramp says AI Spend can pull daily usage and billing data, compare cost per token across providers and models, flag anomalies, and avoid ingesting prompt content. That gives finance one screen for variable AI costs that would otherwise live across provider dashboards and billing exports.
-
It also sets up a services layer on top. Ramp has launched Applied AI Solutions, a team that embeds engineers with enterprise finance organizations to wire together internal systems and build custom agents. That mirrors Ramp's broader bet that the winning vendor will own both the spend data and the workflow layer that decides where AI dollars go next.
From here, Ramp is moving toward becoming the budget allocator for enterprise intelligence spend. As token pricing gets more complex and AI usage shifts from fixed contracts to volatile consumption, the company that can connect provider usage, procurement, accounting, and policy in one workflow will have the inside track to own the next finance stack.