Hi everyone!
I'm Julie VerHage-Greenberg with This Week in Fintech and welcome back to Season 4 of our OG series.
This time I’ve brought in a co-host, my friend Lauren Crossett, Head of Go-to-Market at Spade, a data and AI platform that turns messy transaction strings into structured, verified records.
What started as an idea back in 2022 (when my now three-year-old was still an infant) has grown into one of our favorite annual traditions: spotlighting the people who've been building, studying, and shaping fintech from the very beginning.
This season, we’re kicking off with two guests who bring a rare combination of behavioral science and product rigor: Angela Hung, a behavioral economist who has studied household finance at Carnegie Mellon, Robinhood, and Earnin, and Dan Egan, who has spent 13+ years at Betterment applying decision science to how everyday people invest and save.
In this conversation, we explore what's surprised them most about fintech's evolution, where AI is genuinely changing things versus where it's mostly hype, the limits of "self-driving money," the rise of prediction markets, and what a career high and low actually look like when you've been in this industry long enough to have both.
Let's dive in!
🎧 Episode Summary: Behavioral Finance, Self-Driving Money & the Limits of AI Trust
[00:00 – 05:30] Welcome to Season 4 & Guest Introductions Julie kicks off Season 4 with some warm pre-show energy, introducing Lauren as this season's co-host. Lauren gives her background, GTM leader at Spade, former Quovo, Plaid, and Pinwheel, with a fintech career spanning over a decade. Julie then welcomes Angela and Dan, noting she and Dan go way back (she's visited the old 23rd Street Flatiron office), while Angela is a first-time meeting. Angela introduces herself as a behavioral economist who studied household finance in academia before pivoting to fintech, with stints at Robinhood and earned wage access startup Ernan. Dan jokes that Angela "stole 30% of his lines" before giving his own background: a master's in decision science, seven years at Barclays in London, and 13+ years at Betterment, where he joined when the company was 12 people in a Soho apartment.
[05:30 – 12:00] What Does "Fintech OG" Actually Mean? Julie reflects on the arc of the series and what it means to have been in fintech for 12–17 years. Dan notes he successfully completed a master's in decision science in 2005 and "successfully failed" to get into a PhD program, which sent him toward real-world applications instead. He describes the through-line of his career: the conviction that behavioral insights designed for high-net-worth clients at traditional banks could and should be made available to everyone via technology. Angela shares a similar path, founding a Center for Financial and Economic Decision Making at Carnegie Mellon, then transitioning to fintech after feeling the pull to study how real products were actually being built. The group also bonds over the shared chaos of pivoting careers in early 2020, right as the pandemic hit.
[12:00 – 17:00] Surprises to the Upside, and the Downside Julie asks each guest: what surprised you most about where fintech ended up by 2026?
Angela's upside: The breadth of access. Fintech genuinely brought financial products and services to populations that traditional institutions never reached, from the unbanked and underbanked to first-time investors through Robinhood. Robo-advising democratized advice once reserved for wealthy clients. She calls this the best thing fintech has delivered.
Dan's upside: Affordability alongside access. He points to how dramatically the cost of investing has dropped, from round-trip trades costing 7% in the 1990s to effectively zero today, and the corresponding leap in quality of service for those low fees.
Dan's downside: The limits of delegation. He had hoped consumers would embrace "self-driving money" far more than they have, the ability to link accounts, auto-move excess cash into higher-yield vehicles, and let technology optimize their financial lives. But people hit a wall. They were uncomfortable fully delegating, and it was also hard to communicate the value of sophisticated strategies (like tax-loss harvesting or asset location) to someone glancing at an app on the subway.
[16:30 – 22:00] "Driverless Money" & the Psychology of AI Adoption Lauren picks up the "driverless money" thread, a phrase she feels the industry forgot about but that may be coming back with AI. The group debates how quickly consumers will trust apps to make financial decisions for them.
Angela draws a parallel to self-driving cars: people still want their hands on the wheel, or at least a button they can press to take back control. A hybrid model, automate the decisions, but preserve the opt-out, may be the path that works best behaviorally.
Dan sees a more fundamental psychological shift happening: the accelerating pace of change is making it harder for people (especially younger generations) to think long-term or invest in skills, because the landscape that worked for the generation before them may not apply to them. He frames this as a kind of "hyperinflation of human capital."
On the AI opportunity: Dan is genuinely excited about AI systems that accumulate personal context over time, knowing your family situation, your goals, your financial trade-offs, and becoming a better financial thinking partner than any search engine ever could.
[22:00 – 28:30] What Betterment Actually Looks Like in an AI World Julie asks Dan directly: what does Betterment look like in five years?
Dan is candid that five years may be past the "event horizon" of predictable change. Near-term, he expects the biggest AI-driven shifts to come not at the consumer level but in the advisor and 401k spaces, professional environments where people are already using productivity tools and can adopt AI more naturally. For HR professionals managing 401k plans, AI-generated nudges ("these three employees haven't enrolled yet") will feel seamless in a way they might not for a consumer on their phone on a Saturday morning.
On the consumer side, Dan makes an important distinction: Betterment is a fiduciary. The advice has to be right, deterministic, not probabilistic. AI can explain the advice in personalized, jargon-adjusted language (different for an experienced investor than a first-timer), but it can't be the thing making the call. That division of labor, AI for communication, humans and rules for the actual decision, is where he sees things heading. This feature, a personalized account recommender with AI-written explanations tailored to investor experience level, was literally two weeks old at the time of recording.
[28:30 – 34:00] The Data Layer Underneath Everything Julie flags the connection between AI quality and data quality, noting that Lauren's company, Spade, is doing the unglamorous but essential work of turning messy transaction data into structured, verified records. Lauren picks up the thread: raw connectivity (what Plaid and Quovo pioneered) is only the beginning. Adding merchant context to first-party transaction data is what unlocks personalization that's actually useful, and that consumers can benefit from without necessarily knowing AI is involved.
Angela connects this to a concern about personalization as a concept: it can sound like a euphemism for marketing. But the real promise is proactive financial health, a banking app that spots signs of overspending or falling behind on payments before a crisis hits, and surfaces solutions quietly, before things snowball.
[34:00 – 38:30] What Actually Scares Them About AI The conversation turns to risks.
Dan: Voice spoofing. He recently received a scam call where the same words came through in two different accents, an AI voice spoofer that didn't quite work. But it's close. For anyone who's put their voice on the internet (including podcasters), the spoofability risk is real and growing.
Angela: Two things. First, hallucinations in financial contexts are genuinely dangerous, she cites a friend who was told by an AI platform to hold up to 50% of their portfolio in gold. Second, and more insidiously: learned helplessness. As financial decisions get increasingly outsourced to automation, people may stop paying attention entirely, just as people on autopay miss price increases or forgotten subscriptions. The opt-out exists, but if no one exercises it, the safety net disappears.
Lauren: Connects this to the autopay behavior she's observed, the way automation changes our relationship to our own financial decisions, sometimes in ways we don't notice until something goes wrong.
[34:30 – 39:00] Prediction Markets: Entertainment or Illusion? Lauren asks Angela, given her Robinhood background, for her take on the rise of prediction markets. Angela's honest answer: she doesn't know them deeply, but her core concern is misidentification. Participating for entertainment? Fine. But if someone mistakes a prediction market for a source of genuine insight or an investment vehicle, that's where the harm starts.
Dan goes deeper: he references the Good Judgment Project and Philip Tetlock's superforecaster research, long-running work on who can actually predict world events months or years out. What's changed with crypto-based prediction markets is anonymity plus the ability to profit, which has shifted participation away from genuine long-horizon forecasters and toward insiders making last-minute trades on information others don't have. His conclusion: these markets may self-destruct, because rational counterparties will eventually stop wanting to be someone else's patsy.
[39:00 – 47:05] Career Highs and Lows Julie closes with a fan-favorite question: one high, one low, from a career long enough to have both.
Dan's low (and maybe high): Being removed from a management role leading a team of eight. At the time it felt like failure. A week later, it felt like liberation, 20 fewer meetings a week, more time for the work he actually loves. He's now an individual contributor who also engineers on Betterment's mobile app, which he taught himself during the pandemic. He frames it as a lesson that the traditional ladder isn't the only measure of progress.
Angela's high: Building the research team at Ernan, a small but genuinely interdisciplinary group of four, blending economists, sociologists, an epidemiologist-turned-UX-researcher, and a seasoned qualitative researcher. The collaborative, curious culture they built together and the direct connection to product decisions made it one of the most rewarding professional experiences of her career.
Angela's low: The steep learning curve of translating academic research into fintech product decisions. Coming from academia, she had to completely rewire how she did research (faster, more focused, more tightly tied to specific decisions) and how she communicated it (listening first to understand what product teams actually needed, not just reporting findings).
Dan's high: Tax Impact Preview. Betterment's feature that shows users the estimated tax consequence of a proposed portfolio change before they make it, novel because traditional brokerages have zero incentive to slow you down from trading. The experiment results were striking: roughly 90% of users who saw a meaningful tax impact either reduced their planned change or abandoned it entirely. Dan calls it one of the clearest examples he's seen of behavioral design actually working at scale to help people make better financial decisions. He's still proud of it.
/
