Ep # 169: AI for Dispute Fighting: Turning Usage Data Into Bank-Readable Evidence ft. Patrick Petteruti (Founder, Chargeblast)

Patrick Petteruti 00:02 One kind of process we have in place is a massive log of all the chargebacks that we're seeing across our entire portfolio. If we see any discrepancy or influx, whether it be high chargeback numbers, a lot of unmatched alerts, or erroneous data that we're sending that doesn't look right, we're going to reach out to that merchant. Intro 00:24 Welcome to Across the Funnel, where we dig into concrete Go-To-Market moves across sales, customer success, and account management, so you can build revenue that lasts. Brought to you by Hyperengage and Dextrego. Adil Saleh 00:40 Good morning, everybody. This is the old Hyperengage podcast, and today we are going to drill down on the fintech side of things. With this AI evolution, this is the biggest shift we've had in the past six years. Conversational AI and vertical agents are now shaping how consumer-based industries handle support, especially tier one and tier two, and how they integrate buying processes across workflows. This is a change we didn’t previously dig into deeply with new-age tech companies, specifically fintech startups. Today, we have Patrick, who is the co-founder of one of the finest fintech startups we’ve found and embedded. It took us six months to identify teams doing this at real scale early on. Scale matters when you talk about payment volumes, conversions, and integrations handled daily, weekly, and monthly. In this vertical, it’s about doing things right at scale. Unlike many B2B SaaS startups where you can A/B test and validate quickly, here you must get the product right from day one. I’m talking about Chargeblast, one of the finest fintech startups out of New York City. Patrick is the co-founder. Thank you very much, Patrick, for taking the time. Patrick Petteruti 02:21 Thank you so much for having me. Looking forward to the conversation. Adil Saleh 02:25 Foreign payments have been one of the biggest concerns for fintech and retail. I hear many stories of phones getting stolen, bank apps being accessed, retail transactions happening, e-commerce gift cards being purchased. This ease of penetration has become one of the biggest problems in consumer apps. For context, Chargeblast is serving over 7 billion dollars in annual payment volume with more than 4,000 merchants. We’re talking about real scale and mitigating risk for retail, e-commerce, and DTC businesses. Tell us about the inception. You can’t build this without living the problem yourself as a founder. What did the early journey look like, especially given how engineering-heavy and regulatory-heavy this space is? Patrick Petteruti 03:44 That’s exactly right. To get into chargebacks and fraud prevention, you need firsthand experience with the problem. About three years ago, my two co-founders and I started a consumer AI essay-writing app for students called Cramly.ai. We marketed heavily on TikTok and Instagram using short-form content and scaled to millions of users per month. This was before the big ChatGPT wave, a few months before it launched publicly. We leveraged OpenAI’s backend APIs and accepted payments through Stripe. It was a 20-dollar monthly subscription. As we scaled, we noticed an increase in chargebacks from students using their parents’ cards or accounts like Cash App. They used the product and then filed disputes saying they didn’t buy it or weren’t satisfied. We often lost those disputes on Stripe. A key issue we learned was that if your monthly chargebacks exceed around 1 percent of transactions, that crosses the acquiring bank’s risk threshold. Stripe can shut off your ability to process payments, forcing you onto high-risk processors charging six to seven percent per transaction. We met with every fraud and chargeback mitigation platform trying to solve this and realized there was no good turnkey solution integrated directly with Stripe. Most AI consumer apps are built on Stripe because it’s low-code and easy to use, but there was no effective way to reduce dispute rates. This led us to our core product, chargeback alerts, which act as a notification layer between issuing banks and merchants. When a customer initiates a dispute, we receive an alert before it flows through card networks to Stripe. That allows the merchant to issue a refund before the chargeback hits. You turn chargebacks into refunds, keep dispute rates low, and continue processing on Stripe. That’s how Chargeblast started and how we found our first customers by targeting Cramly.ai-like merchants. Adil Saleh 07:34 As a SaaS founder, I relate deeply. Stripe alone isn’t enough. You need a strong security layer alongside it. Managing chargebacks under one or two percent is a massive challenge, especially as you scale. Thinking Stripe-first, how did you expand to work with other payment gateways? How does that workflow look for companies not exclusively on Stripe? Patrick Petteruti 08:34 We started Stripe-first. About 90 percent of our first thousand customers were on Stripe. But enterprises don’t rely on one processor. They may have Adyen, Braintree, Airwallex, and others. Today, we integrate with over 35 processors, gateways, and orchestrators. The key is matching issuer data to transaction data. Alerts from something like Apple Pay are tokenized and look different from what’s stored in Stripe. Our value comes from accurately matching that data so merchants can act on it. That’s how we help reduce chargeback rates as much as possible. Adil Saleh 10:09 This has a direct revenue impact. From Series A to Series C startups, financial reporting and dispute management become board-level conversations. From a B2B SaaS perspective, what does onboarding look like? How easy is integration, and how does it fit into workflows? Patrick Petteruti 10:52 Fast onboarding was a major growth driver. Historically, this took 45 to 60 days with manual processes. Our onboarding takes about 10 to 15 minutes. Merchants either install our app from the Stripe marketplace or provide an API key for processors like Adyen or Braintree. It’s fully self-service. No sales call required. Adil Saleh 12:05 Many SaaS companies want to move off Stripe due to high transaction fees but fear the risk. How do you view that tradeoff? Patrick Petteruti 12:59 Stripe is easy to use and has strong authorization rates due to excellent bank relationships. That’s especially valuable in medium to high-risk verticals. It is expensive, though. We help merchants diversify volume across processors, improve acceptance rates, and optimize fees through payment orchestration. Routing transactions by geography or risk profile can significantly improve outcomes. Adil Saleh 14:54 You’re expanding rapidly into DTC and e-commerce. What are the most common preventable disputes you see? Patrick Petteruti 15:27 E-commerce subscriptions are big. Supplements, nutraceuticals, telehealth products like GLP-1 consultations often generate disputes when prescriptions aren’t issued. Another major vertical is iOS apps moving off Apple’s native payments to Stripe following the Apple vs Fortnite ruling. Apple takes 30 percent per transaction, so many apps now need chargeback protection on Stripe. We also support high-risk industries like iGaming, sweepstakes, social casinos, gift cards, eSIMs, travel rewards, and data providers like VIN lookups. Recurring billing is always more prone to disputes. Adil Saleh 17:32 In B2B SaaS, overpromising during sales can lead to disputes. Do you see that often? Patrick Petteruti 18:40 Not heavily in traditional B2B SaaS. Where we do help is automating chargeback fighting using AI. We submit usage data, IP matching, and evidence back to issuers. Most B2B clients we work with are marketplaces or merchant-of-record platforms. Chargebacks aren’t typically a massive issue in pure B2B software. Adil Saleh 19:46 What about crowdfunding platforms and lifetime deal marketplaces? Patrick Petteruti 20:20 Not much. What we do see often is pre-launch sales. Accepting payments before product delivery creates a huge influx of chargebacks, especially in software and e-commerce. Adil Saleh 20:54 Do you work with Shopify, Amazon, or eBay? Patrick Petteruti 21:01 We work with Shopify. Amazon and eBay handle payments internally, so we can’t integrate. Our Shopify integration mirrors our Stripe functionality. Adil Saleh 21:30 Customer education seems critical. Many merchants don’t realize the problem until it’s too late. Patrick Petteruti 22:18 Exactly. Many merchants come to us after crossing the point of no return. Education is critical but difficult because this issue stays invisible until accounts are flagged. We focus heavily on awareness through LinkedIn and Twitter, explaining dispute rates and why merchants get removed from Stripe. Stripe isn’t acting arbitrarily; they’re passing on bank-imposed risk. We also work directly with processors to re-enable accounts by demonstrating reduced risk and sub-0.1 percent dispute rates. Adil Saleh 24:10 When you say processors, you mean Stripe, Shopify, and similar platforms? Patrick Petteruti 24:14 Exactly. Stripe, Shopify, Adyen, Airwallex. We work closely with Stripe’s risk and sales teams to help remove reserves and keep merchants processing. Adil Saleh 25:17 Let’s talk Go-To-Market. Education is harder than penetration, but this is a painkiller, not a vitamin. How are you scaling? Patrick Petteruti 26:22 We started with founder-led, guerrilla sales through Twitter. One viral tweet alone drove massive early traction. Today, our Go-To-Market has three channels: direct merchant sales, partnerships, and referrals. We partner with Stripe, Adyen, and platforms like sticky.io. Referrals are now our biggest channel. I get about 20 warm intros per week from existing clients. Patrick Petteruti 29:48 That’s just volume back in Stripe’s pockets. The other avenue, like I mentioned, is referrals. The merchants we support in niche verticals, such as iOS apps, operate in tight networks where everyone talks to each other. The same applies to dropshipping e-commerce. Today, our biggest channel is referrals from merchants we already support. If they know someone running a similar business, they send them to Chargeblast because they trust our ability to handle it and provide strong support. That’s created a snowball effect. I get about 20 introductions per week just from existing clients. Adil Saleh 29:39 Amazing. It really comes down to running multiple growth avenues simultaneously as a founder. How large is your post-sales team? Patrick Petteruti 29:48 On the sales side, including myself, my CEO, and four to five others, we’re around seven people. We also have a large customer support team in the Philippines. At this point, it’s over 100 team members. They also handle inbound sales. If someone comes to our website and uses live chat asking about Chargeblast, they handle onboarding, which is almost a sales role. We’re distributed globally. Adil Saleh 30:27 So post-sales is close to 100 people. Is there a strong education component internally? We talked earlier about education being critical. With a team that large, how much are you investing in training, enablement, and tooling? Patrick Petteruti 31:09 It depends on the role, but our Philippines team handles onboarding and understands the product fundamentals deeply. We add new hires directly into a real Chargeblast account so they see how everything works in practice. We maintain Notion documentation explaining each product and its purpose. Our alerts product reduces dispute rates, and we also have products that fight disputes and deflect disputes. The team is well-equipped to onboard and support merchants, so leadership stays fairly hands-off. White-glove support is a big part of our value. Adil Saleh 32:11 One of the biggest challenges I see is helping merchants forecast disputes before they get shut down. If one or two more disputes come in, what actions should they take? Doing this at scale with 100 people is difficult. Do you have systems, triggers, or recommendations to help merchants act proactively? Patrick Petteruti 33:29 That’s something we’re constantly improving. One process we have is a massive log of all chargebacks across our entire portfolio. We track descriptor data and dispute trends. Because we integrate directly into processing, we can see dispute rates rising in real time. If we see discrepancies, sudden spikes, unmatched alerts, or incorrect data being sent, we proactively reach out to the merchant. Our support team follows a clear SOP for this. We also offer 24/7 live chat on our website and web app, with an average response time under 60 seconds. If something goes wrong, merchants speak to someone almost immediately. Adil Saleh 34:40 I like that approach. AI works well for tier-one support, like what Intercom does, but for mission-critical issues, you need humans. When margins are thin, this can’t be fully automated. Looking ahead, as you scale beyond 5,000 merchants, how are you thinking about education and development long-term, especially with AI? Patrick Petteruti 36:35 Interestingly, our first hire in the Philippines was originally for Cramly.ai, and they’re still with us at Chargeblast. We focus heavily on retaining strong talent. Rather than constantly adding headcount, we want to deeply educate existing team members because the product knowledge is very specialized. This isn’t easy to onboard quickly. Experienced team members train new hires. We also maintain an active Slack channel where we share product updates, industry changes, and things affecting Stripe or underlying data. Keeping the support team informed is critical, and it’s an ongoing challenge as we scale. Adil Saleh 38:23 That makes sense. The problem is massive across industries like e-commerce, healthcare, and gaming. To wrap up, looking toward 2026, with AI agents and automation evolving quickly, what excites you most about AI’s role in chargebacks and fraud? Patrick Petteruti 40:09 The biggest opportunity is on the dispute-fighting side of the chargeback process. That’s where merchants submit documentation to issuers, and a bank agent decides whether the merchant recovers the funds. We’re building AI models that automatically gather usage data and customer information from wherever merchants store it and compile a clean, comprehensive report for bank reviewers. Adil Saleh 41:02 Even moving the win rate to 50 or 60 percent by improving how information is submitted through Stripe or banks like Chase would be huge. Patrick Petteruti 41:11 Exactly. Right now, it’s manual. A bank agent reads an eight-page report and decides in a few minutes. AI-generated reports must be extremely clear, accurate, and structured with key points upfront so they’re easy to evaluate. Adil Saleh 42:43 So it has to be highly specialized, not just a generic summary. Patrick Petteruti 42:51 Exactly. We’re working on automating data ingestion from wherever merchants store usage data. For example, car washes might rely on license plate data, while restaurants might use photo or video evidence of purchases. Adil Saleh 43:01 So it’s about pulling data from wherever it lives and managing automated submission and storage. Patrick Petteruti 43:51 Exactly. Some merchants use Zendesk, others have custom systems. That’s why integrations matter so much. Outro 43:56 Thank you for listening to Across the Funnel. If you got one useful Go-To-Market idea from today’s episode, share it with a teammate and hit follow. Explore Hyperengage at hyperengage.io and Dextego at dextego.com.