The Payments Experts Podcast

Taming Chargebacks With Real AI Agentics | Special Industry Insider ChargeFlow & Ben Herut | PEP078

• Expert Payments Attorneys of Global Legal Law Firm • Episode 78

Share episode

Taming Chargebacks With Real AI Agentics

Too many merchants breathe easy under legacy chargeback ratios, only to be blindsided when their PSP tightens the screws. We sit down with Ben Herut, co-founder at ChargeFlow, (https://www.chargeflow.io/) to unpack the shift toward post-payment risk and why Visa’s VAMP is forcing acquirers and processors to act earlier—and harder—on portfolio-level exposure. If you rely on card rails for growth, this conversation shows how to protect revenue without clobbering conversions.

We trace Ben Harut’s path from engineering to payments risk and dig into how ChargeFlow uses post-payment data, alerts, and AI-driven workflows to cut chargebacks and protect revenue. We also break down Visa’s VAMP, why PSP thresholds change the game, and how merchants should respond.

• career path from engineering to payments risk
• bank-side underwriting, KYC, fraud and chargebacks
• founding in high-risk and lessons learned
• what ChargeFlow does post-payment and pre-fulfillment
• risk scoring using cross-merchant and outcome data
• handling alerts including TC40 and RDR
• strategies to refund or fight disputes
• AI agents for representment and QA feedback loops
• what VAMP changes for acquirers and PSP thresholds
• portfolio-level risk, BIN pressure, and early enforcement
• why proactive prevention protects processing access

We start with Ben’s journey from electronics engineering to bank-side risk, through launching an EMI in the high-risk space. That experience shapes a practical view: fraud prevention cannot end at authorization. ChargeFlow focuses on the critical window after approval and before fulfillment, where merchants can use post-payment data, cross-merchant signals, and scheme alerts like TC40 and RDR to flag risky orders, request verification, or cancel before losses mount. For digital goods and financial products, we explore how delayed access flows and behavioral patterns unlock smarter decisions than blanket declines.

When disputes hit, evidence wins. Ben explains how AI agents compress months of training into days, assembling compliant, precise representments and feeding results back into models. The goal isn’t buzzword AI; it’s a genetic workflow with guardrails, explainability, and QA loops that cut manual work and raise win rates. We also compare refund-first versus fight-first strategies, and where high-value transactions justify the extra effort.

Then we tackle VAMP’s impact. Even “safe” merchants can trigger portfolio pressure at the BIN level, prompting PSPs to hold payouts, request mitigation plans, or offboard accounts. Understanding tighter PSP thresholds, modeling risk appetite, and staying current with card scheme changes are now core operating skills. Whether you sell physical products or digital access, the path forward is clear: centralize post-payment risk scoring, handle alerts with discipline, and standardize dispute workflows so your team can focus on growth.

If this episode helps clarify your approach to chargebacks, subscribe, share with your ops or finance team, and leave a quick review—what’s your stance today: refund or fight?

**Matters discussed are all opinions and do not constitute legal advice.  All events or likeness to real people and events is a coincidence.**

Visit Global Legal Law Firm today:
https://www.globallegallawfirm.com/podcasts/

A payments podcast of Global Legal Law Firm

SPEAKER_03:

So what about uh I'm sure you're you've you know you're probably actually excited about the VAMP. You might be one of the few people that are actually looking forward to this. There's your real advanced sales force. Uh but what are your thoughts on, you know, now that we're into it, what are your thoughts on Vamp overall?

SPEAKER_01:

Welcome to the Payments Experts Podcast, a podcast of Global Legal Law Firm. We hope you enjoyed this episode. Really excited today. We've got in studio joining us, a managing partner of the law firm, James Cooper, as well as our special guest joining us from Israel, uh Ben Harut of Chartflow. You can find Ben over at Chartflow.io. Ben, we're really excited for you to make this podcast with us. We get the fun topic of talking about your business and chargebacks and that whole world. Welcome.

SPEAKER_00:

Thank you so much. Thanks for having me.

SPEAKER_03:

Yeah, Ben, why don't you tell us? I mean, I'm always curious about how people find themselves in this industry. Because once you're in, it's really no getting out. It's like the crypts. Yeah.

SPEAKER_00:

That's true. So actually, it's a funny story, right? We talked about uh previously uh about how we got into the business. And to be honest, I started as an electronics engineer. So I worked for about three years in the electronics firm, R D, and then uh uh I moved into analytics, and then I I went to uh honeymoon with my wife in Mexico, and then she saw on Instagram a post by a friend saying that she just relocated to uh London. And my wife said, you know what? It sounds interesting to live in Europe. Now, me is as a child, I lived for six years in Paris, so I I was kind of accustomed to living in Europe, so you know what, let's give it a try. So we actually started looking for opportunity, and then I found one as a risk analyst in an acquire bank in Germany.

SPEAKER_03:

So your wife saw someone on Instagram and got jealous and was like, we're moving halfway around the country.

SPEAKER_00:

Yeah, yeah, exactly. Um, and this is how it started essentially. Then once, as you mentioned, once it went in, there's no going back.

SPEAKER_03:

There's no going back. What did you so you were a risk analyst? What were you doing at the bank specifically, generally?

SPEAKER_00:

Well, everything to be honest, from uh you know, monitoring risk from merchants and doing compliance and uh and and underwriting and uh KYC and whatnot, everything that had to be done. Um yeah, I did it for about two years. Then I moved to an issuer bank in Germany as well, N26. It's an Austro-German bank. Um, then I also yeah, managed the charge backs and card fraud prevention teams, and I just uh continued from there.

SPEAKER_03:

And you realize like a lot of people after you've worked at the bank for a few years, and you're working with all these people and you're going, well, they're making a whole lot more money than I am.

SPEAKER_00:

Yeah, I mean, to be to be honest, one of my uh latest positions before this one, the before the current one, I actually opened an EMI in the Netherlands with a partner, and it was in a high-risk space. And this is where you do actually make the real money.

SPEAKER_03:

Right. But it's risky. But it's risky, yeah, high risk, high reward. Are you running cross-border with the US?

SPEAKER_00:

Um, yeah, so we currently at Charles we do have uh a team situated in New York. They've got an office, we've got some uh people in the in Canada and Europe as well, and the main headquarters are in Israel. So we do actually all over the place. I try to get as much into uh conferences, uh Europe and the US, and just you know, getting together with the payment experts whenever I can. Why don't you tell us about ChargeFlow then? Right. So essentially ChargeFlow helps merchants recover lost revenue and reduce charge complexity. Um the idea is that we understand the domain itself and we've developed an internal tool that uses a genetic workflow, but real AI, not like everybody likes to use this buzzword. So I essentially part of the time I actually prompt engineering some agents, and we are able to uh you know provide a hands-free solution for merchants um that uh deal with a chargeback issue. And to be honest, anyone who uh uses cards for the payments has a chargeback issue.

SPEAKER_03:

Right, right. So you're is it is it helping with the the dispute the chargeback? Is it helping trigger reversals? Is it you know stopping them before it happens?

SPEAKER_00:

Yeah, so so all of the above. Essentially, we come in uh post-payment, right? We've got the whole prepayment set up, you've got different uh fraud prevention firms, but we go just after the payment was approved, and essentially before it was uh um, let's say fulfilled, if we're talking about uh physical goods, then before the goods were shipped. So we have a very unique uh product that we've just developed that's sort of like uh aggregates uh risk signals to essentially tell the merchants whether a payment was risky or not, and then it gives them a chance to either cancel it or uh require additional verification. Then we move on to the alerts, and by alerts, I mean TC40, ATOCA, CDLN, RDR, and all this good stuff. And then if you know everything uh still passes through, then we've got the charge recovery uh product. So yeah, any anything after the payment was made.

SPEAKER_03:

How do you deal? So, you know, I I can appreciate it as you know, before the goods are sold, but how do you deal with a company that's selling like a uh financial product? So let's say like a real estate investment company of you know, here I'll process a million dollars on my debit card to invest in this real estate. How would you help someone like that?

SPEAKER_00:

So essentially it really depends on the merchant itself. If they have some sort of a mechanism that requires some time before the payment is actually processed, and then uh and between the the sh providing access to whatever the non-physical good is, then we are able to step in and we have some sort of a consortium of data from all of our merchants, and we process uh a little over 15,000 merchants, so we've got a big knowledge base and data that we can utilize, and then we are able to uh incorporate both behavioral signals, just data-related signals, as well as chargeback signals, uh, which is something that mostly lacking in the fraud prevention firms. They do not see the after the payment was done. So we take all of this, we throw it into a model that we've uh developed internally, and then we are able to generate some sort of uh uh risk score plus uh explainability version or layer on top of it.

SPEAKER_03:

So, in this situation, where the the merchant would you know take this money and they might go, oh hey, hold on. This uh this looks a little iffy. Why don't you uh double check and uh maybe don't don't authorize it?

SPEAKER_00:

Exactly, exactly, because essentially using the cards, you leave traces, right? You you use your cards in different places, not just with the real estate agent. So if this other payments were captured by our environment, then we are able to get some signals from the other purchases that were made. Uh, this is how how we are able to sort of like cross-function uh the data from different sources. Um, and we see it quite a lot. Uh, I just had a conversation uh last week with uh with a um social gaming uh merchant, and they mentioned the fact that they tried to use some sort of uh product or similar product, but the outcome was that all of their customers were risky. Well, it's a risk, high-risk merchant, right? Right. So essentially you have to some sort of like tweak the model to accommodate the the risk level of the industry, and this is something that we are that we achieved.

SPEAKER_03:

Yeah, that's great. I mean, I guess that you know there's no perfect solution, and there's always people that get through. How do you deal with the people that get through? Like if I'm using your company and I'm going, oh hey, you're supposed to tell me now that I guess it's a million-dollar chargeback.

SPEAKER_00:

Well, it really depends on the risk appetite of the merchant itself, right? Some merchants would like to, you know what, just refund everything which is risky, anything that we get an RDR for or something, just refund it and let it bund let it be done with. And some merchants have the appetite to, you know what, let's try to fight these the chargebacks because they're high value. So it's really up to the preference of the merchant itself.

SPEAKER_03:

And then do your does your company help with fighting the chargebacks too?

SPEAKER_00:

Yeah, definitely. So what we did is we developed like an internal mechanism that uses a genetic workflow. Essentially, I've trained in the past so many live human agents, and I would say typically it takes about six months until someone is qualified enough to really handle a chargeback. But then you can do the same thing with an AI agent within, let's say, two, three weeks, to be honest. So it's I mean, it doesn't yes, you do have to take care of hallucination, but if you prepare a good rug and a good knowledge base and you're able to QAid along the way and so so have some sort of a feedback loop of the results, you're able to maintain a machine that is somewhat human, but doesn't do human mistakes.

SPEAKER_03:

I'm sure you're you're you know, you're probably actually excited about the VAMP. Might be one of the few people that are actually looking forward to this. That's your real advanced Salesforce. Uh but what are your thoughts on, you know, now that we're into it, what are your thoughts on Vamp overall?

SPEAKER_00:

Right. So Vamp as any Visa announcement is is uh is something that which is rolling. It's a it's a live beast that changes and you know adapts and so on. And usually when they close the the first the initial announcement, then nobody understands it, then they do some recalibration, then still nobody understands it. It's a back and forth. Um but essentially the idea is that now acquires and processors are gonna be accountable um for managing the risk, the portfolio. And one of the biggest mistakes that I see merchant doing is saying, you know what? My thresholds are now so high. I'm I I I'm not even close to a thousand chargebacks, right? Or TC15 and TC40. I'm not even close to 1.5%. So I'm fine, nobody cares. But then the reality strikes, and then essentially, since the PSP itself has a much lower threshold, like 0.3, 0.2, it means that any merchant which is higher than 0.2, 0.3 is gonna negatively impact the PSP itself. Which means that without any written regulation, the PSP might you know block out payouts, right? Close accounts, just do risk uh assessments and so on. So it's it's like uh yeah, merchants need to understand that even though they're not close to the ratio, they are gonna be impacted.

SPEAKER_03:

That's right. Yeah, I mean, I think that's the that's the scary thing because I talk to you know big ISO owners and they're going, oh, we're low risk, what do we care? And I'm like, Do you know which bank you are at? They might always well it's what it doesn't matter, it's a you know any of the banks because the it's at the bin level. So um I think it's way more problematic. But to your point, they'll roll it out if it blows up like everybody, like it I guarantee it will if it's enforced in its current you know, current structure, then they'll just change it.

unknown:

Right.

SPEAKER_00:

So I essentially started on the first of October, the enforcement period, and I've already received letters from PSPs to merchants saying, you know what, you guys are too risky. Provide us with a mitigation plan, even though they're nowhere in in you know close to the threshold.

SPEAKER_03:

Right, right. Yeah, you just want that good stuff. So um we uh I feel like I've learned enough. What else do you want to tell us about charge flow? You get the you get the last word.

SPEAKER_00:

Okay, so I mean, essentially, as I mentioned, charge flow, we've built like a universal solution for merchants and the PSPs or the platform that supposed uh support those merchants. Um chargebacks are manual, chargebacks are confusing, and chargebacks are not the main business of a merchant, right? Right. They should focus on selling whatever they sell, not on doing chargebacks. I always give this uh example. If you're sick, you're gonna go to the doctor. If you have a lawsuit, you're gonna go to a lawyer. You're not gonna try to represent your. I mean, you can, but yeah, you know how the outcomes be. So we all we do is we do chargebacks uh recovery and prevention. Um, so the idea is that we are on top of all the changes. You mentioned Vamp, there are so many changes. Every six months or so, the card schemes are uh having a different announcement coming out. Um, so we get this all into our models and we try to provide like a full hands-free solution for merchants. There they could be involved if they want it, it's fully customized. But uh, the idea is that we just take this burden off and let them focus on actually revenue uh production.

SPEAKER_03:

All right, great. Well, thanks for coming on. We like to keep these short because at the end, Jeremy always says, Thank you for listening to this. Go on and on and on. Um but yeah, thanks for coming on. I think uh your product is definitely something that people can benefit from and then also be a little bit scared of. Yeah, hopefully. Both thanks so much for having me.

SPEAKER_00:

It was uh great pleasure.

SPEAKER_03:

All right, thanks, Ben.

SPEAKER_01:

Thanks so much, Ben. And as James mentioned, thank you for listening this long to the Payments Experts Podcast, a podcast of Global Legal Law Firm. Please go visit our special guest, Ben Harut, over at chargeflow.io. All the information down below. And as always, we will see you on the next one. Bye-bye. Thank you for listening to this episode of the Payments Experts Podcast, a podcast of Global Legal Law Firm. Visit us online today at Global Legalaw.com. Matters discussed are all opinions that do not constitute legal advice. All events or likeness to real people and events is a coincidence.