Research · October 31, 2026

Casap's $56M Series B — AI Agents That Have to Prove a Dispute, Not Just Process It

Casap raised a $56 million Series B led by Base10 Partners, with Experian and Synchrony as strategic investors, to bring AI agents to the full dispute lifecycle for financial institutions.

What happened

On October 28, 2026, Casap, an AI agent platform for financial services, announced a $56 million Series B funding round led by Base10 Partners, with strategic participation from Experian and Synchrony. Existing investors including Emergence Capital and Lightspeed Venture Partners continued their support. The round was oversubscribed, the company says, and brings its total funding to $90 million — just over a year after a $25 million Series A. Over that same period, Casap says its revenue grew 10 times, more than double its original target. No absolute revenue figure is disclosed, so the "10x" is a company-reported growth multiple, not a revenue number.

Casap says it will use the funding to advance its AI platform and help financial institutions achieve outcomes that were previously too expensive, complex, or time-consuming to deliver at scale. The expansion target is explicit: beyond its starting point in disputes, into fraud, collections, claims, and complex exceptions — the high-stakes operations where fragmented systems and manual investigations currently produce financial loss, delays, and poor customer experiences.

The job the agents actually do

Casap's agents are not a chatbot layer on top of a call centre. They work the full dispute lifecycle — intake, investigation, decisioning, evidence collection, recovery, and consumer communication — inside the systems financial institutions already use. They apply the institution's own policies and procedures, bring people in when a decision needs human judgment, and leave every decision supported by evidence and a clear record of the actions taken, so outcomes are explainable and auditable.

That auditability is the product, not a feature. A dispute is a regulated, adversarial process: one side says a charge was not theirs or not as described, the other side must prove otherwise, and the burden of proof is what makes disputes expensive. Casap's starting insight, in CEO and co-founder Shanthi Shanmugam's words, was that disputes were "one of the hardest problems in financial services" — and therefore the right place to prove that AI could change outcomes, not just speed up work. "Today we're cutting fraud losses in half while resolving disputes in days instead of months," she said. "Now our customers are asking us to bring that same transformation to the rest of their operations."

The company-reported results, across more than 100 banks, credit unions, fintechs, and processors, are:

All five figures are company-reported, aggregated across customers, and presented here as such. No per-customer breakdown, no time window for the totals, and no independent audit of the figures is published in the announcement.

Why the strategic investors matter as much as the lead

A $56 million Series B led by a venture firm is ordinary. The strategic pairing is the more informative part of this round.

Experian's role, as described in the announcement, is to explore a new generation of first-party fraud and consumer-intent signals — combining Casap's view of transaction behaviour and fraud with Experian's identity and risk expertise, to help the industry distinguish genuine fraud from first-party misuse, especially as agentic commerce increases. That is a data partnership dressed as an investment: the signal that separates an honest dispute from a dishonest one gets better with more transaction and identity context, and Experian is one of the few parties that can supply it at scale.

Synchrony's role is distribution and credibility with a different kind of institution. Synchrony, a consumer financing company serving over 100 million Americans alongside major brands and hundreds of thousands of small and midsize businesses, frames the investment as a bet on customer service at the dispute moment. "When a customer disputes a charge, it's an important moment to get right," said Kelly Shaw, SVP at Synchrony Ventures. "Casap's agentic AI helps resolve disputes faster and more accurately while stopping first-party fraud, along with service teams' oversight."

Base10's Adeyemi Ajao frames the thesis in outcome terms rather than cost terms: "Casap has shown that agentic AI technology can change that tradeoff. The same technology lowers the cost of complex work, while improving speed, quality, and consistency. Disputes proved that model in one of the hardest environments in financial services, and we believe the opportunity now is much larger."

Casap was founded by Shanmugam (formerly at Robinhood) and Saisi Peters (formerly at Chime). Emergence Capital's Lotti Siniscalco adds a build-method detail worth noting: the team spent time with the employees buried in manual dispute work before writing code, and designed the product so staff do not need to become technologists. For a tool that must be adopted inside regulated operations teams, that is a go-to-market claim as much as a product claim.

What this does not show

The announcement discloses no valuation, no absolute revenue, no customer names, and no pricing or margin figures. The 10x revenue growth, the 51% fraud-loss reduction, the 96% win rate, and the 73% automation share are all company-reported aggregates with no methodology, no customer-level distribution, and no statement of how many of the 100-plus institutions are paying at scale versus piloting. A win rate "compared with a 60% industry average" is only as strong as that average's definition, which the announcement does not give.

The expansion question is also open. Disputes are a bounded, evidence-heavy workflow with a clear decision at the end — a good fit for auditable agents. Collections, claims, and complex exceptions each carry different regulations, different counterparties, and different definitions of a good outcome. The announcement asserts the same transformation will transfer; it offers no evidence from those categories yet.

What would turn this round from a thesis into evidence is the hard version of the numbers: win rates and resolution times broken out by institution type and case type, the share of the 100-plus institutions still active a year later, and a first collections or claims deployment with the same five metrics reported the same way.

Sources