Freehand Raises $75M Series B for Autonomous Spend Agents

There is a slow, unglamorous corner of enterprise software, verifying invoices, chasing vendor disputes, and reconciling contracts against what actually got billed, that has resisted automation for decades. It is also, quietly, where a lot of money leaks out of large companies. Freehand's pitch is that agents can now do that job well enough to plug the leak, and enterprise buyers are apparently willing to bet on it: the San Francisco startup just raised a $75 million Series B, according to Crunchbase News, co-led by Battery Ventures and NewRoad Capital Partners with participation from Nexus Venture Partners and former U.S. Commerce Secretary Penny Pritzker. That brings total funding to $100 million on top of a $25 million Series A raised in March 2024.

The customer list is what makes this more than another agentic-AI announcement. Freehand says it counts roughly 50 customers including Meta, Johnson & Johnson, Pfizer and Cardinal Health, and CEO and co-founder Nitin Jayakrishnan told Crunchbase News the platform processes billions in payments across 60 to 70 countries and hundreds of currencies without human supervision. The agents reportedly read contracts, policies, emails and internal data to check whether an invoice matches what was actually agreed, then negotiate adjustments directly with the vendor. Jayakrishnan and Abhijeet Manohar, who co-founded and recently sold their previous company Pando, started Freehand in February 2024.

Why this matters if you do not run a procurement team: this is one of the first large-company AI deployments where the agent is not drafting a memo or summarizing a meeting but actually moving money and holding the pen in a vendor conversation. Jayakrishnan described the case bluntly, saying an invoice can arrive from a supplier claiming 'you owe me $16.948 million for everything that I've done for you in the last six months,' with few proof points to check whether the number is right. The company's own claim is that it recovers 5% to 10% of total spend and cuts procure-to-pay cycle times by more than 70%.

The honest caveat is that those numbers are Freehand's, not audited, and none of the named customers have publicly quantified what they have saved. What the reporting does not give you is a per-transaction error rate, a pricing model, or how much of the flow still needs a human in the loop on high-value invoices. Autonomous negotiation with a strategic supplier is also the kind of thing that can go wrong quietly, over months, before anyone notices.

Still, the direction is worth watching. If agentic accounts-payable and category management actually work at Meta and Pfizer scale, the incumbents in enterprise procurement software are the ones with something to lose.