Casap Raises $56M to Scale Agentic AI Across Financial Services

Direct answer: Casap, an AI agent platform built for financial services, has raised $56 million in a Series B funding round led by Base10 Partners, with strategic backing from Experian and Synchrony. The oversubscribed round brings the New York company’s total funding to $90 million, just over a year after its $25 million Series A, and it will be used to push its agents beyond card disputes into fraud, collections, claims and other complex financial operations. For UK small businesses, the story matters less as a product to buy and more as a signal: the back-office work that decides whether a disputed payment is refunded, a fraud claim is believed, or a collection is chased is being rebuilt around agents that can investigate, decide and act, with a human still in the loop for the hard calls.

What happened

Casap announced the round on 28 October 2026. Base10 Partners led it. Experian and Synchrony joined as strategic investors, a telling combination: one is a giant of credit and identity data, the other a major consumer finance company that works with hundreds of thousands of small and mid-sized businesses in the United States. Existing investors Emergence Capital and Lightspeed Venture Partners also continued their support. Casap did not disclose a valuation.

The context for the raise is unusually concrete for an AI funding story. Casap says its revenue has grown tenfold since the Series A, more than double its original target, and that more than 100 banks, credit unions, fintechs and payment processors now use its platform. Those are company-reported figures, and they should be read as such, but the shape of the story is consistent: a product that started in one painful, narrow workflow and is now being pulled by customers into adjacent ones.

That starting workflow was transaction disputes. Anyone who has run a small business knows the shape of it. A customer challenges a card payment. The bank or payment provider has to work out whether the charge was genuine, whether the goods or service were delivered, whether the cardholder is telling the truth, and what the card scheme rules say about who bears the loss. Done by hand, that investigation is slow and expensive. It involves pulling records from several systems, checking location and spending patterns, assembling evidence, writing to the customer, and meeting deadlines that, if missed, can decide the case by default.

Casap’s pitch is that this is exactly the kind of work an AI agent is suited to, provided it is built with guardrails rather than given a free hand. Its agents work through the full lifecycle of a dispute, from intake and investigation through decision-making, evidence collection, recovery and customer communication. They apply the institution’s own policies and procedures, not a generic set of rules, and they escalate decisions that need human judgement. Every action is backed by an auditable record of evidence and reasoning, so the institution can explain, to a regulator or a court if necessary, why it reached a particular outcome.

The numbers Casap publishes for that dispute work are striking, and they are worth quoting precisely because they set the bar the company will now be judged against. Across its customers, it reports up to a 51% reduction in first-party fraud losses, a 96% chargeback win rate compared with an industry average of around 60%, resolution times cut from as long as 90 days to roughly 8 days, 73% of dispute cases resolved through agentic automation while retaining human oversight, and human processing time per case falling from about 45 minutes to 1–2 minutes.

First-party fraud deserves a short explanation, because it is central to why this round matters and why it is different from the card-fraud stories of a decade ago. Traditional card fraud is committed by a criminal using someone else’s card. First-party fraud, sometimes called friendly fraud, is committed by the cardholder themselves: a legitimate purchase is disputed as if it were not made, or a refund is claimed for goods that were in fact received. Casap says cardholders now account for close to a third of fraudulent card transactions, a shift it links to AI tools that make filing a dispute easier and to social media groups that share tips for getting goods for free. Whether that share is exactly right is hard for an outsider to verify, but the direction of travel is widely recognised across payments: when disputing a charge becomes a few taps in an app, more disputes arrive, genuine and otherwise, and the cost of investigating each one does not fall on its own.

That cost problem explains a detail that will sound familiar to any small business owner who has ever lost a small dispute. Many banks, Casap and others in payments have noted, simply approve every dispute below a certain amount, often around $30, because investigating costs more than the claim is worth. In some cases the cutoff is much higher. The result is a quiet subsidy for dishonest claims, paid for by honest merchants and, ultimately, by honest customers through higher prices and fees. An agent that can investigate a $30 claim for a fraction of the human cost changes that arithmetic, which is why the chargeback win rate and the resolution time matter more than the raw funding figure.

How the technology actually works

It helps to be precise about what Casap means by an agent, because the word is now used for everything from a chatbot to a fully autonomous system. Casap’s agents are not a single model answering questions. They are a platform that combines specialised agents with financial-services expertise and with teams that work alongside institutions to redesign their operations before the agents go live.

In practice, the workflow looks like this. A dispute arrives through the institution’s existing channels. An agent gathers the relevant evidence: transaction records, location data, spending patterns, delivery information, prior communication with the customer, and the institution’s own policy for that type of case. It weighs that evidence against the policy, decides whether the case can be resolved automatically, and either acts, for example by preparing and submitting the evidence package for a chargeback, or escalates it to a person with a summary of what it found and what it recommends.

Two design choices matter here. The first is that the agents work across the systems the institution already uses, rather than requiring a rip-and-replace of core banking software, which is one of the slowest and riskiest projects any financial firm can undertake. The second is that the agents are constrained by institution-specific rules and controls. A bank does not hand Casap a blank cheque to decide cases however it likes; it sets the policies, the thresholds for escalation, and the areas where a human must decide. That is slower to deploy than a generic chatbot, and it is also the reason a regulated institution can plausibly defend the outcome.

Casap also talks about its work in terms of outcomes rather than efficiency, and that distinction is worth taking seriously. Efficiency is doing the same job faster and cheaper. An outcome change is a job that previously could not be done at all becoming possible: a small dispute investigated properly instead of written off, a fraud pattern spotted across cases that no single investigator would have connected, a customer in financial difficulty offered a workable plan instead of being passed between departments. Casap’s chief executive, Shanthi Shanmugam, who co-founded the company with Saisi Peters, frames the ambition in exactly those terms, arguing that if the agentic era only delivers the same outcomes faster, it will have missed the point.

The company’s background is relevant to that claim. Shanmugam previously worked in product roles at Robinhood, Meta and Salesforce, while Peters is a veteran of Chime, the US consumer fintech. Casap was founded in 2023, which makes its growth to more than 100 institutional customers in about three years fast by financial-services standards, where sales cycles are long and procurement is cautious. The presence of Experian and Synchrony as strategic investors, rather than purely financial backers, also suggests something about where the company expects its next growth to come from: identity and risk data on one side, and the messy reality of consumer credit and retail commerce on the other.

Comparison: how Casap sits against its rivals

Casap is not the only company applying AI agents to financial back-office work, and the comparison helps UK readers judge how distinctive it really is. The broad field splits into roughly three groups.

The first group is the large, established fraud and risk vendors, including the card schemes themselves and companies that have sold rules-based fraud detection for decades. Their strength is scale and data: they see transactions across thousands of institutions and can spot patterns no single bank could. Their weakness, historically, has been that their tools flag cases for humans to investigate rather than investigating themselves. Casap’s bet is that the investigation itself, not just the flagging, can now be automated safely.

The second group is the general-purpose agent platforms, the companies building agents that can work across any industry, from customer service to procurement. Their strength is flexibility and speed of deployment. Their weakness in financial services is depth: a generic agent does not know card scheme rules, chargeback reason codes, or the evidence a particular issuer needs to win a particular type of dispute. Casap’s counter-argument is that depth is the product, and that a platform built specifically for high-stakes financial operations will outperform a generalist in the cases where a wrong decision costs real money.

The third group is the in-house route: large banks building their own agent teams. The biggest institutions can afford this, and some are doing it seriously. But most banks, credit unions and payment firms cannot hire and retain the specialist talent, and they cannot spread the cost of building dispute, fraud, collections and claims agents across a large enough base. That is the gap a company like Casap is trying to occupy: specialist depth, sold as a platform, to institutions that are too small to build it themselves but too regulated to buy a generic tool.

There is also a geographic comparison worth making. The United States, where Casap operates, has a dispute and chargeback system shaped by its own card scheme rules and consumer protection laws. The UK system is different in important ways: the chargeback scheme rules still apply, but UK consumers also have Section 75 of the Consumer Credit Act for credit card purchases over £100, the Financial Ombudsman Service as a backstop, and a regulatory culture, through the Financial Conduct Authority, that places heavy emphasis on treating customers fairly and on being able to explain automated decisions. An agent that works well in the US cannot simply be switched on in the UK. It would need to be reconfigured for UK rules, UK evidence standards and UK escalation paths. That is a real barrier to entry, and also a real opportunity for whoever gets it right first.

What it means for UK small businesses

Most UK small businesses will never sign a contract with Casap, and that is fine. The effects will arrive indirectly, through the banks, payment providers and platforms they already use, and through the competitive pressure Casap’s growth puts on those providers. Here is how, in practical terms.

First, faster and fairer dispute outcomes. If you sell online, disputes are a cost of doing business, and they are a cost that falls unevenly. A large retailer can absorb a run of chargebacks and staff a team to fight them. A small retailer, a tradesperson who takes card payments, or a subscription business with a few hundred customers cannot. When a payment provider’s dispute process improves, because its agents can gather evidence quickly and present it properly, small merchants win twice: genuine disputes are resolved faster, which protects customer relationships, and dishonest disputes are more likely to be challenged, which protects margins. The difference between a 60% win rate and a 96% win rate, if those figures hold in practice, is not an abstraction. It is money that stays in a small business instead of being written off.

Second, fraud costs that currently hide in prices. First-party fraud is not a victimless accounting entry. When a merchant loses a chargeback on goods that were delivered, the merchant loses the goods, the shipping cost and the payment processing fee. When that happens at scale, merchants raise prices, tighten refund policies, or stop selling certain products online altogether. A system that can distinguish genuine fraud from first-party misuse more accurately, which is exactly what Experian’s involvement hints at, combining Casap’s transaction behaviour insights with identity and risk data, could allow merchants to be more generous with genuine customers while being firmer with dishonest ones. That is a better outcome for everyone except the people gaming the system.

Third, the collections and claims angle. Casap says it is expanding from disputes into fraud, collections, claims and complex exceptions. For a small business, collections is often the most uncomfortable part of the job: chasing late payment without damaging a relationship you may need next month. If agents can handle the routine chasing, the evidence gathering and the payment-plan administration, while escalating genuinely difficult cases to a person, small firms could recover more of what they are owed with less awkwardness. The same logic applies to insurance claims and supplier disputes, where the paperwork is the barrier rather than the merits of the case. This is the same pattern we have seen in other parts of small-business automation: the tool does not replace judgement, it removes the administrative weight that stops judgement being exercised promptly. Readers who already automate payment chasing will recognise the principle from tools covered in our guide to automated payment reminders for UK small businesses, and the same discipline, clear policies, clean records, human escalation, applies here at a larger scale.

Fourth, customer experience as a competitive edge. Small businesses often compete on service rather than price, and nothing damages service faster than a customer who feels ignored during a dispute. A resolution in 8 days instead of 90 days is not just an operational metric; it is the difference between a customer who stays and a customer who leaves a one-star review and never returns. As banks and payment providers adopt faster dispute handling, customers will come to expect it everywhere, including from the small firms they buy from directly. Firms that can match that speed, by keeping clean transaction records, responding to evidence requests promptly, and using tools that surface the right information quickly, will have an advantage over firms that treat disputes as an afterthought.

Fifth, a warning about expectations. Agents make the back office faster, but they do not make the underlying rules simpler. UK small businesses still need to understand their own obligations: when a customer is entitled to a refund, what evidence a chargeback requires, how Section 75 works, and when to escalate to a human at their payment provider rather than accepting an automated outcome. The firms that benefit most from agentic tools will be the ones that already have tidy records and clear policies, because an agent can only act well on good data. Firms with messy records will find that automation amplifies the mess rather than fixing it. That is a useful discipline to apply before buying any AI tool, and it connects directly to the broader question of what AI actually costs a small business once implementation, data clean-up and oversight are counted, which we examined in our guide to how much AI costs UK small businesses in 2026.

There is also a sector point. Financial services is one of the first industries where agentic AI is being trusted with decisions that move money, because the industry has the data, the audit requirements and the economic incentive to get it right. What works there tends to spread: first to insurance and professional services, then to the operational tools that small businesses use every day, from bookkeeping to customer service. Watching Casap is therefore also a way of watching what the next generation of small-business software will look like: less dashboard, more delegation, with a human setting the rules and checking the exceptions.

Honest limits / what to watch

A funding announcement is not a product review, and this one should be read with the usual caution. Several limits and open questions deserve attention.

The performance figures are company-reported. The 51% fraud reduction, the 96% win rate, the 8-day resolution time and the 73% automation rate are Casap’s own numbers, drawn from its own customers, and they have not been independently audited in public. They are plausible, and they are consistent with what better investigation should deliver, but they are not a guarantee that any particular bank, let alone any particular small merchant, will see the same results. The honest test will be whether those figures survive as Casap moves into new workflows, collections and claims are different problems from disputes, with different rules and different failure modes, and as it serves institutions with messier data and less mature processes.

Regulation is the second watchpoint. Agents that make financial decisions sit squarely in the territory regulators care about: explainability, bias, data protection and accountability. In the UK, the FCA expects firms to be able to explain automated decisions, to test for unfair outcomes, and to maintain human oversight where it matters. In the EU, the AI Act adds further obligations for high-risk systems. Casap’s emphasis on auditable records and human escalation is clearly designed for this environment, but the regulatory treatment of agentic systems is still evolving, and a single high-profile failure, an agent wrongly freezing an account, or wrongly denying a genuine claim, could slow adoption across the sector.

The third watchpoint is concentration and dependence. If a growing share of the industry’s dispute and fraud decisions run through a small number of agent platforms, those platforms become systemically important in a quiet way. A bug, a data breach or a change in a platform’s policies could affect many institutions at once. Small businesses, who have no direct relationship with the platform, would feel the effects without having a say in the remedy. That is not a reason to reject the technology, but it is a reason to value providers, and regulators, who insist on transparency, portability and clear lines of accountability.

Fourth, the UK question remains open. Casap has not announced UK operations, UK partnerships or UK regulatory approvals. Its current customers are described in general terms as banks, credit unions, fintechs and processors, without a geographic breakdown. UK readers should therefore treat the direct effects as medium-term rather than immediate, and the indirect effects, through global payment providers and through competitive pressure, as the nearer-term channel. If Casap or a rival announces a UK launch, that will be a story in its own right, and the details of how it adapts to Section 75, the Ombudsman and FCA expectations will matter more than the funding figure.

Finally, there is the human question. Casap says human processing time per case falls from 45 minutes to 1–2 minutes, which is presented as a benefit, and in many ways it is: fewer hours spent on repetitive investigation, more time for complex cases and customer relationships. But it also implies a workforce change inside financial institutions, and the quality of that change depends on how it is managed. If staff are redeployed to harder, more rewarding work, everyone gains. If they are simply cut and the remaining staff are left supervising agents they do not fully understand, quality will suffer in ways that may not show up in headline metrics for a while. Small businesses should watch not just the technology, but the service quality of the institutions they rely on.

Key facts

Timeline

• 2023: Casap is founded by Shanthi Shanmugam and Saisi Peters, with a focus on transaction disputes, one of the hardest operational problems in financial services.
• 2025: Casap raises a $25 million Series A, backed by investors including Emergence Capital and Lightspeed Venture Partners.
• 2025–2026: The platform expands from disputes into broader financial operations as customers ask its agents to take on more work, and revenue grows tenfold on a company-reported basis.
• 28 October 2026: Casap announces a $56 million Series B led by Base10 Partners, with strategic participation from Experian and Synchrony, bringing total funding to $90 million.
• Next: Watch for expansion into collections, claims and complex exceptions at scale, for any UK or European market entry, and for regulatory responses to agents making financial decisions.

What to watch next

Three things will tell us whether this round was a milestone or just a milestone on a press release. The first is customer expansion: does Casap add institutions outside its current base, and does it do so in new geographies, particularly the UK and Europe, where the regulatory bar is different? The second is workflow expansion: do the results seen in disputes hold up in collections and claims, where the human and legal stakes are different? The third is the strategic investors: do Experian and Synchrony move from investors to deep product partners, combining identity, risk and consumer finance data with Casap’s agents in ways that are visible to merchants and customers, or does the strategic label remain mostly financial?

For UK small businesses, the practical takeaway is simpler. You do not need to buy an agentic platform to benefit from this shift, but you do need to be ready for it. Keep clean transaction and delivery records, respond quickly when a payment provider asks for evidence, review your refund and dispute policies so they are clear and fair, and treat dispute handling as part of customer service rather than as an administrative nuisance. The firms that do that will be the ones that gain most when the back office gets faster, whichever platform ends up powering it.

Sources: Casap’s Series B announcement via PR Newswire, 28 October 2026; TipRanks and Dealroom coverage of the round; Base10 Partners’ announcement. All performance figures in this article are company-reported unless stated otherwise.

FAQs

What is Casap?

Casap is a New York-based AI agent platform for financial services. Its specialised agents investigate transaction disputes, fraud cases, collections and claims, make decisions under each institution’s own policies, and take action across the systems a bank or fintech already uses, with human oversight for judgement-sensitive decisions.

How much did Casap raise in its Series B?

Casap raised $56 million in an oversubscribed Series B led by Base10 Partners, with strategic participation from Experian and Synchrony and continued support from Emergence Capital and Lightspeed Venture Partners. The round brings Casap’s total funding to $90 million, just over a year after its $25 million Series A.

What results does Casap claim for its customers?

Across more than 100 financial institutions, Casap reports up to a 51% reduction in first-party fraud losses, a 96% chargeback win rate against a roughly 60% industry average, dispute resolution in around 8 days instead of up to 90 days, and 73% of dispute cases resolved through agentic automation with human oversight retained.

What does Casap’s funding mean for UK small businesses?

Most UK small businesses will not buy Casap directly, but they will feel its effects through faster dispute outcomes, fewer wrongly refused refunds, and banks and payment providers that can afford to investigate smaller claims properly. The same agentic approach is also spreading to the tools small firms use for invoicing, collections and customer service.

Is Casap’s AI replacing people in banks?

No, not in the way Casap describes it. Its agents handle routine investigation, evidence gathering and execution, and escalate judgement-sensitive decisions to human staff. Every decision is backed by an auditable record of evidence and reasoning, which matters in a regulated industry where a bank must be able to explain why it acted.

What should UK firms watch next from Casap?

Watch whether Casap expands beyond US institutions into the UK and Europe, how regulators respond to agents making financial decisions, and whether the claimed fraud and resolution figures hold up as the platform moves from disputes into collections, claims and other high-stakes operations.

CTA: Get a Free AI Consultation