On June 4, Ramp closed a $750 million Series F at a $44 billion valuation. That number tells one story. The product tells a different, more important one.

The round was led by ICONIQ, GIC, and Ontario Teachers’ Pension Plan, with Goldman Sachs Alternatives, D.E. Shaw, Morgan Stanley Investment Management, and Insight Partners all coming in as new investors. These are not venture tourists chasing hype. They are the kind of institutions that show up when a business has crossed into something durable.

And Ramp has crossed. The company now serves more than 70,000 customers, processes over $200 billion in annualized purchase volume, and crossed $1 billion in annualized revenue. It reached positive free cash flow. Its customer list includes Visa, Uber, Shopify, Anduril, and Figma.

But the number that actually matters for anyone running a business is this: median Ramp customers are saving 50 percent more dollars and 32 percent more hours per year compared to a year ago. That is not a feature. That is operational leverage at scale.

What the Agents Actually Do

Ramp has been quietly building a set of finance agents that are now running in production for thousands of teams.

The Policy Agent enforces spend rules with 99 percent accuracy and catches 15 times more out-of-policy spend than non-AI alternatives. The AP Agent handles accounts payable processing. The Accounting Agent manages reconciliation and financial close.

These are not AI assistants that answer questions. They are autonomous systems that review transactions, flag violations, match invoices, and close books without waiting for a human to initiate each step.

Ramp has also deepened its partnership with Visa to let AI agents execute autonomous corporate payments with real-time controls. That is the logical endpoint of agentic finance: the agent does not just suggest the payment, it makes it, within guardrails the business has defined.

On the roadmap: procurement agents, vendor-onboarding agents, budgeting agents that track against plans in real time, and reconciliation agents that auto-match transactions and flag mismatches before month end.

The CEO’s stated view is that fully autonomous finance is two to three years away. Based on the trajectory, that estimate looks conservative.

Why This Round Is a Signal, Not Just a Story

Ramp grew its total payment volume by 170 percent year-over-year in March 2026, its highest growth rate in three years, despite being 20 times the size of the business that achieved those early growth numbers. That is the kind of figure that does not happen without genuine product-market fit at enterprise scale.

The investors in this round are not betting on a category that might emerge. They are betting that Ramp has already won enough of it to make the position defensible.

What This Means for Business Owners

The shift underway in finance operations mirrors what is happening in customer service, operations, and IT. Routine, rule-based work is moving to agents. The humans running finance teams are moving into strategy, exception handling, and oversight.

For most businesses, this is a good thing. Finance teams have spent decades doing work that is fundamentally information processing: checking whether transactions match policies, reconciling accounts, chasing receipts. AI agents do all of it faster and with fewer errors.

The practical question for business owners is not whether to use AI in finance. It is which parts of the finance function to hand off first, what guardrails to set, and how to build the internal capability to manage a system that increasingly manages itself.

Getting the sequencing right matters. Dropping agents into a finance function that lacks clean processes and clear policy documentation tends to produce noise, not efficiency. The companies getting the most out of agentic tools are the ones who defined their processes before they automated them.

That is a pattern we see consistently across the businesses we work with. The AI does not fix unclear processes. It scales them.

Enterprise DNA put together a free field guide on exactly this: the full Claude ecosystem, Claude Code, and how to roll agents out without breaking things. Get the guide.