Mike de Vere runs Zest AI, a company that has been applying machine learning to credit underwriting for over two decades, starting with some of the largest banks on the planet and now serving a large share of the credit union market. Since his last appearance on the show three years ago, Zest has expanded well past underwriting into fraud detection and portfolio management, tied together by an intelligence layer and a generative AI companion called LuLu. Mike makes a specific argument in this conversation: machine learning still makes the credit decision, generative AI makes the feedback loop faster, and the real advantage available to community financial institutions is a willingness to pool what they know.

What We Covered

• Zest today, from underwriting to fraud to portfolio management
• Why the intelligence layer is what makes an ecosystem
• Starting with Discover, Citi and Freddie Mac, then moving down market
• LuLu, named after a corgi, and what she actually does
• Safety and soundness as the first use case for most institutions
• Replacing quarterly reports that used to take weeks
• Peer benchmarking versus building your own data lake
• Collective intelligence across 2,000 credit models in production
• Why generative AI has no role in making the credit decision
• Shrinking model refit cycles from 18 months to daily evaluation
• Zest customers versus non-customers on growth, delinquency and efficiency
• Cash flow underwriting, and why generic national models fail
• Zest Protect and fighting AI-powered fraud with AI
• The two objections that come up most in sales conversations
• Takeaways from the IQ AI Lending Forum in Santa Fe

Key Takeaways

• The performance gap is measurable. Comparing Zest customers to non-customers across 2024 and 2025, Mike says his customers grew 16 times faster, ran roughly 20 points lower on delinquency, and were 501 basis points better on efficiency ratio.
• Generative AI belongs around the credit decision, not inside it. Zest still uses supervised, locked-down machine learning models for underwriting, because a regulator will ask you to explain the decision. What generative AI changes is the speed of evaluation, from an 18-month refit cycle to daily.
• Comparison is where the value sits. A lender looking only at its own data lake has visibility on itself and nothing else. LuLu is built to normalize performance data across institutions so a chief lending officer's instinct can be checked against thousands of real policy instances rather than one career's worth of experience.
• Community lenders have a structural advantage they underuse. The credit union industry holds roughly $2.4 trillion in assets. If it acted as one institution, it would be bigger than Wells Fargo, and unlike the big banks these institutions are actually willing to share.

About Mike de Vere

Mike de Vere is the CEO of Zest AI, the AI lending technology company that has been doing machine learning in credit since well before AI became a standard fintech conference track. He came to Zest from a career in data and consumer insights, with leadership roles at J.D. Power, The Harris Poll and Nielsen. Zest now touches $5.6 trillion in assets under management, and by the end of this year expects one in three credit union members to have their consumer loans decisioned with its technology.

Connect with Fintech One-on-One:

• Tweet me @PeterRenton
• Connect with me on LinkedIn
• Find previous Fintech One-on-One episodes

Nandan Sheth has spent 25 years in payments, building three growth companies along the way, including Harbor Payments (sold to American Express) and Acculynk (sold to First Data/Fiserv). He now runs Splitit, which takes a different path than most buy now, pay later providers: instead of originating a new loan, it turns the credit a consumer already has on their existing card into an installment plan, with no underwriting, no social security number, and no new debit card for repayments. With agentic commerce infrastructure being built in real time, Nandan argues that a frictionless installment option is exactly what merchants need to avoid being commoditized on price inside an LLM shopping platform.

What We Covered

• Three growth companies across 25 years in payments
• What attracted Nandan to Splitit from Fiserv
• Card-linked installments with no underwriting or new loan
• The card loyalist versus the credit needy
• $3.5 trillion of unused credit sitting on US cards
• Merchant-funded 0% economics and where the budget comes from
• A $1,300 average order value versus $250 to $300 for standard BNPL
• Point of sale through the Samsung Wallet integration
• Backing Google's Universal Commerce Protocol
• The overlooked small business to large supplier B2B use case
• Chargebacks, repudiation, and who carries the risk in agent-led purchases
• Splitit Go for the face-to-face services economy

Key Takeaways

• BNPL is really two markets, not one. Card loyalists want rewards, protections, and habit, while the credit needy want a new line of credit. Nandan thinks both get served, but by different products.
• The economics work because the merchant treats it as marketing spend. About 98% of Splitit's volume is a merchant-funded 0% plan, priced comparably to a percentage-off promotion, and it lifts average order value roughly four times over standard BNPL.
• In agentic commerce, price and delivery speed are the easiest things for an LLM to compare. A 0% installment option gives merchants a third lever that is not pure price competition.
• The B2B version may be the stronger use case. Small business owners face both a time problem and a working capital problem, which is a sharper reason to hand off buying to an agent than a consumer shopping for a polo shirt.

About Nandan Sheth

Nandan Sheth is the CEO of Splitit, the card-linked installments platform. He moved to the US from the UK 25 years ago and has spent his entire career in payments and fintech, including running e-commerce and omni-channel commerce at Fiserv. He previously built Harbor Payments, acquired by American Express, and Acculynk, acquired by First Data/Fiserv.

Connect with Fintech One-on-One:

• Tweet me @PeterRenton
• Connect with me on LinkedIn
• Find previous Fintech One-on-One episodes

Escheatment is a $70 billion problem hiding in plain sight: every state, territory, and dozens of countries have laws that hand dormant and unclaimed accounts over to the government after three to five years of inactivity. Allen Osgood, co-founder and CEO of Eisen, left a five-and-a-half-year run as a payments product manager at Coinbase to build the compliance infrastructure that helps banks, brokerages, and crypto platforms reunite customers with their money before the states ever claim it. In this conversation, Allen makes the case that crypto is about to collide with escheatment rules written in the 1960s, and that most institutions have no idea how large their own dormant balances really are.

What We Covered

• What escheatment actually is and how the state-by-state rules work
• The $70 billion states are holding for more than one in seven Americans
• Missingmoney.com and what happens after money is remitted
• Ohio's fight over using unclaimed property to fund a football stadium
• The Walter story: an E-Trade Amazon account liquidated to Delaware
• What counts as a "dormant" account and why logins matter
• Where Eisen plugs into the escheatment process
• Why reactivation beats remittance, and the Binance.US 48% case study
• Why institutions are blind to their largest dormant balances
• The 12-to-24-month gap where accounts just age untouched
• Displacing big-four spreadsheets with a single pane of glass, forecasting, and access controls
• Data volume as the hardest engineering problem, and where AI earns its keep
• The Claims Portal and QR-code reactivation
• Why crypto makes escheatment far more painful, from volatility to dust
• The coming wave of crypto liquidations and the tax problem
• Channel strategy with the cores like Fiserv, and the road to 1099 and tax reporting

Key Takeaways

• The best escheatment outcome is no escheatment at all. Eisen's real value is retention: keeping customers, deposits, and assets in the institution rather than shipping them to the state.
• Institutions routinely underestimate their exposure. One prospect thought it had 10,000 accounts about to escheat, the real number was 100,000. The disconnect sits between the compliance team and the data on the ground.
• Crypto changes the stakes. States generally require liquidation, so a dormant token gets sold, creating an unwanted taxable event and, if the market rips afterward, another Walter waiting to happen.
• Stale data is the enemy. The information that comes due for escheatment is by definition three to five years old, so address enrichment (LexisNexis, Socure, USPS NCOA) and early engagement are what actually move the reactivation numbers.

About Allen Osgood

Allen Osgood is the co-founder and CEO of Eisen, a compliance operations platform that automates escheatment and account offboarding for financial institutions. Before founding Eisen, he spent about five and a half years as a payments product manager at Coinbase, where he first ran into the strange world of unclaimed property and stayed through the company's IPO.

Connect with Fintech One-on-One:

• Tweet me @PeterRenton
• Connect with me on LinkedIn
• Find previous Fintech One-on-One episodes

Accounts payable has produced multiple billion-dollar companies, yet its mirror image, accounts receivable, remains almost entirely manual at most enterprises despite decades of software spend. In this episode, Caitlin Leksana, co-founder and CEO of Fazeshift, explains why AR has remained unsolved and how her company's AI agents are changing that. A mechanical engineer turned BCG consultant turned founder, Caitlin came to the problem the hard way, doing her own AR by hand at a previous startup, and her outsider's view of a stubborn back-office chore is exactly what makes the conversation worth your time.

What We Covered

• A million AR analysts doing manual work in the US
• Why accounts payable got solved and AR did not
• The leverage imbalance between AP and AR departments
• The swivel chair problem and fragmented data
• $200 million in unapplied cash on one balance sheet
• Fazeshift as a context layer, not a rip-and-replace
• Why traditional SaaS and if-then logic could never scale AR
• The collections, cash application, and AR inbox modules
• Human in the loop and building trust when AI touches money
• Training agents on historical data and tribal knowledge
• From Y Combinator to a Series A led by F-Prime
• The vision for the context layer and autonomous finance

Key Takeaways

• AR is the inverse of AP, and every bill is someone else's invoice, so the market is at least as large and mostly uncaptured.
• The real unlock is not the AI model but unifying fragmented data across the ERP, bank, CRM, and inbox into a single context layer.
• Human in the loop with full auditability is what earns risk-averse finance teams' trust, and it is how agents move toward full automation over time.
• Some of the best unsolved startup problems are the ones furthest removed from an engineer, because no one with the tools to fix them ever felt the pain.

About Caitlin Leksana

Caitlin Leksana is the co-founder and CEO of Fazeshift, a San Francisco startup building AI agents for accounts receivable. She earned bachelor's and master's degrees in mechanical engineering from Georgia Tech, advised Fortune 500 companies at BCG, and earned her MBA at Harvard Business School before founding a crypto marketing startup and then Fazeshift. The company went through Y Combinator's Summer 2024 batch, raised a $4M seed led by Gradient Ventures, and announced a Series A led by F-Prime in 2026.

Connect with Fintech One-on-One:

• Tweet me @PeterRenton
• Connect with me on LinkedIn
• Find previous Fintech One-on-One episodes

Sahej Suri is the founder of Blue Dot Investors, a late-stage growth equity firm that invests exclusively in fintech across both primaries and secondaries. Before Blue Dot, he built his career at J.P. Morgan, TPG, and as chief of staff to Nigel Morris at QED Investors. In this conversation, Sahej explains the scrappy origin story of the firm, the overlooked opportunity in fintech secondaries, and his new report with FT Partners on the coming fintech liquidity supercycle, including the finding that the top 100 private fintechs now out-earn the top 100 public ones.

What We Covered

• Sahej's path from J.P. Morgan to TPG to QED
• The 2008 recession and why access to financial services stuck with him
• The happenstance origin story of Blue Dot
• Why fintech is closer to biotech than to generalist tech
• The gap in the market for late-stage fintech specialists
• Why the top 10 names dominate secondary market activity
• Finding undervalued companies outside the marquee names
• The "Liquidity Supercycle" report with FT Partners and how it came together
• Why the top 100 private fintechs out-earn the top 100 public ones
• The state of the IPO window and the SpaceX bellwether
• Why the 2025 IPO cohort cleared a much higher bar
• The have versus have-nots dynamic in fintech fundraising
• The Blue Dot dinner series and building community
• His AI thesis and where the value creation will land
• A 10-year view on fintech as an asset class

Key Takeaways

• The best fintech companies are now private, and on the top 100 they out-earn their public peers on revenue, a finding Sahej says had never been put on paper before.
• Fintech rewards specialists. Banking, payments, capital markets, and insurance are almost different worlds, and most investors who piled in during 2021 without that depth are no longer around.
• The IPO window is real but conditional. The 2025 cohort was roughly three times the size on revenue and more profitable than historical norms, and the near-term window hinges on how bellwether listings perform.
• Sahej's bet on AI value creation is not the startups or the large AI labs, but the scaled fintechs that already own distribution and customer trust.

About Sahej Suri

Sahej Suri is the founder and Managing Partner of Blue Dot Investors, a New York-based late-stage growth equity firm investing exclusively in fintech across primaries and secondaries. He previously worked at J.P. Morgan in the financial institutions group, at TPG in growth equity and buyouts, and as chief of staff to Nigel Morris at QED Investors. Blue Dot came out of stealth in early 2026 and manages roughly $100M in assets, with a team of six and around 30 advisors. Peter is an advisor to Blue Dot Investors.

Connect with Fintech One-on-One:

• Tweet me @PeterRenton
• Connect with me on LinkedIn
• Find previous Fintech One-on-One episodes

Dimitri Masin was one of the first 30 employees at Monzo, where he led AI and data science as the bank grew from 30 to 4,000 people. That vantage point showed him where the real work in financial services still lives: the manual, repetitive customer operations running behind the app. In 2023, he co-founded Gradient Labs to automate that work with fully autonomous AI agents, and the company now serves more than 30 fintech and financial services customers. In this conversation, we get into why co-pilots can quietly degrade quality and compliance, why Dimitri believes full autonomy is the safer path, and the story behind what may be the largest known AI agent deployment in banking.

What We Covered

• From Google to one of the first 30 people at Monzo
• The second half of the fintech transformation
• Why customer operations never got reinvented
• What GPT-4 unlocked at the start of 2023
• Putting banks on autopilot
• Sitting as an orchestration layer over existing systems
• The 15% customer experience uplift over human teams
• Why cost savings are more nuanced than people expect
• How bank implementations and bake-offs actually work
• Why co-pilots can degrade quality and compliance
• The case for full autonomy over a human in the loop
• Benchmarking agents against the human team, not perfection
• Redeploying staff instead of cutting headcount
• The largest known AI agent deployment in banking
• Why banks aren't seeing productivity gains yet
• The build-it-ourselves mindset shift
• A five to ten year view of the transformation
• How the US bake-off culture plays to a specialist's advantage

Key Takeaways

• The overlooked opportunity in banking is not the app experience but the manual operational work behind it: customer support, AML, fraud, KYC, onboarding, and screening.
• Co-pilots can backfire. When suggestions are right 90% of the time, people start accepting them blindly, which degrades quality and compliance in the other 10%.
• No agent is correct 100% of the time, and that is the wrong bar. The right question is whether the system beats the human team it replaces, which becomes the benchmark.
• Automation has not meant layoffs at any of Gradient Labs' customers. Teams get redeployed to complex, higher-empathy work like vulnerability and financial difficulty cases.
• The bottleneck on transformation is not the technology, which has existed since GPT-4, but how slowly organizations diffuse and adopt it. Dimitri's horizon is five to ten years.

About Dimitri Masin

Dimitri Masin is the CEO and co-founder of Gradient Labs, a London-based startup building autonomous AI agents that run customer operations for regulated financial services companies. Before founding the company in 2023 with two former Monzo colleagues, he was among the first 30 employees at Monzo, where he led AI, data science, financial crime, and fraud as the bank scaled to roughly 4,000 people. He started his career at Google.

Connect with Fintech One-on-One:

• Tweet me @PeterRenton
• Connect with me on LinkedIn
• Find previous Fintech One-on-One episodes

Yuval Refua is the Chief Product Officer at Navan, the global travel and expense platform he joined seven years ago when it was still just a travel booking service. Since then, he has built out its payments and expense products from the ground up, turning the company policy that used to live in a PDF into code that runs on the card itself. This conversation matters because T&E is one of the most universally disliked workflows in business, and Navan is rethinking it from scratch just as AI and agentic commerce start to reshape how companies spend.

What We Covered

• Falling in love with credit cards at American Express
• Why Navan started as a travel-only booking service
• The reconciliation pain that led to launching a card
• Coding company policy directly onto the card
• Real-time approval the moment you swipe
• Why travel-first beats procurement-first
• Context as the key to managing distributed spend
• Going global with VAT, GST, per diems and mileage
• The e-invoicing wave hitting more countries
• The GTA model for revealing complexity gradually
• The Expense Admin Companion and recommended actions
• From single approvals to bulk to full automation
• The Visa partnership and the Connect product
• Waymo for travelers, Formula One for finance

Key Takeaways

• The expense report exists to answer a question that company policy already settled. Coding that policy onto the card removes the work instead of automating it.
• Starting from travel gives Navan context (where the employee is, why they are there, who they are visiting) that procurement-first tools lack, which makes per-employee limits far smarter.
• Going global is less about features and more about mastering country-by-country tax, e-invoicing, per diem and mileage rules.
• The path to full automation runs through trust. Navan moves finance teams from a single recommended action, to bulk approvals, to hands-off automation, which is also how it intends to handle agentic spend.

About Yuval Refua

Yuval Refua is Chief Product Officer at Navan. He started two companies of his own early in his career before moving into fintech and product management at Thomson Reuters, then American Express, where he developed a deep love for credit cards and the rails behind them. He joined Navan around seven years ago and has built out its payments and expense products from the ground up.

Connect with Fintech One-on-One:

• Tweet me @PeterRenton
• Connect with me on LinkedIn
• Find previous Fintech One-on-One episodes

This episode is part of our occasional Fintech Revealed series, where we do an extended deep dive into one topic with two industry experts. The topic today is vertical fintech, and I am joined by Matt Hennessy, the Business Lead at Increase, the modern banking infrastructure company, and Jamie Fox, the General Manager of Fintech at Tekion, the AI-native cloud platform that runs the entire business for auto dealerships across the US, Canada, and the UK.

Tekion built its embedded banking on Increase, so the two of them give us both sides of the same story: the platform that lives inside the dealership and the infrastructure that connects it to the banking system. We get into the surprisingly large money flows inside a single dealership, why paper checks still beat instant rails for many operators, how compliance and trust get engineered into the product, and just how big this embedded banking opportunity gets.

What We Covered

• What vertical fintech is and why it matters now
• The money flows hiding inside a single car dealership
• Why outbound dealer spend is roughly 2x inbound
• Operating account vs. ledgering account adoption paths
• Dealer-to-dealer payments as a ledger change with zero rail fees
• Instant rails: RTP, FedNow, and Request for Payment
• The persistence of paper checks and the cost to operationalize them
• Direct Fed access vs. layers of middleware
• Compliance as code, codified into the product
• Building trust in building blocks
• Where agentic payments and "know your agent" fit in
• How large the embedded banking opportunity ultimately gets

Key Takeaways

• Owning the financial system of record inside core operating software is the defensible position in an age when light "systems of engagement" can be replicated with AI.
• Outbound payments, not inbound, are the bigger prize: US auto dealerships pushed out roughly $1.3 trillion in 2024, about 2x what they took in.
• The barrier to instant rails is education, not technology. Many dealers do not know RTP or FedNow exists, or that they can pay a vendor any day of the week.
• Trust cannot be launched all at once. Holding a dealer's operating cash is a different level of trust than processing a payment they can fall back on, and it is earned in building blocks.

For the founding story and more about Increase, check out my conversation with CEO and Founder Darragh Buckley from last year.

Connect with Fintech One-on-One:

• Tweet me @PeterRenton
• Connect with me on LinkedIn
• Find previous Fintech One-on-One episodes