Behind the Numbers: an eMarketer Podcast
Episode: Welcome to 2026: AI Agents Manage Money and the Bank-Grade Stablecoin Economy | The Banking & Payments Show
Date: December 30, 2025
Host: Rob Rubin (EMARKETER Head of Advisory)
Guests: Myra Thomas (Senior Analyst, eMarketer), Grace Broadbent (Senior Analyst, eMarketer)
Episode Overview
This episode explores two transformative trends for 2026 in banking and payments:
- The increasing use of agentic AI in consumers' financial lives
- The evolution of stablecoins as bank-grade payment rails
Rob Rubin leads an insightful conversation with analysts Myra Thomas and Grace Broadbent, examining how AI-powered agents are creeping into daily banking and whether stablecoin-backed, programmable money will upend the foundations of how we pay and get paid.
Key Discussion Points & Insights
1. Icebreaker: “Delegate or Don’t You Dare” (01:21–02:36)
- The hosts and guests play a game revealing their willingness to automate various money chores through AI agents.
- Negotiating bills? Delegate.
- Automatically moving cash to savings? Don't you dare.
- Canceling unused subscriptions? Delegate.
- Choosing payment rails? Delegate.
- Insight: People are more likely to delegate tasks they find confusing or uninteresting, but hesitate on core financial decisions involving risk or large sums.
- Quote:
“The things that you would delegate are likely things you're not actually doing today.”
— Rob Rubin (02:36)
- Quote:
2. State of AI in Banking, 2025 (03:17–05:18)
- 70% of banks already use agentic AI in some capacity, but for consumers, it’s mostly in the background.
- Most visible uses today: automation for bank employees and internal ops; consumer-facing solutions are still experimental or piloted.
- Examples:
- Citi and US Bank partnering with Mastercard for AI-driven payments technology, allowing cardholders to link cards to AI agents. (04:34–04:54)
3. AI Adoption: Consumer Trust and Education (05:18–08:11)
- 34% of consumers don't want banks using AI at all.
- Primary concerns:
- Transparency
- Data security and privacy
- Fear of losing control
- Consumer education is essential. Most consumers don't understand AI’s current, non-scary applications (like fraud detection) and have misconceptions due to "AI headlines."
- Quote:
“Banks have been using AI for at least decades, right? ... Fraud would be way more plentiful than it is right now if all banks were not using AI-based fraud protection.”
— Grace Broadbent (06:07)
- Quote:
4. Adoption Hinges on Use Cases, Not Abstract Technology (07:00–09:31)
- For mass adoption, introducing AI agents via specific, helpful tasks (like canceling subscriptions) builds trust and drives usage.
- Cultural/market context matters:
- In markets reliant on digital infrastructure from the start, super apps and agentic payments are more mature.
- US/EU adoption is slower due to legacy systems and ingrained behaviors.
5. Stablecoin Payment Rails: Poised for 'Bank-Grade' Adoption (09:33–11:58)
- 2025 marked a turning point for regulatory support, especially by the Trump administration.
- In 2026, legacy banks begin serious stablecoin adoption, moving the tech from fintech “experimentation” to institutional “product.”
- Major retailers (e.g., Starbucks, Walmart, Dunkin Donuts) are experimenting with their own branded stablecoins.
- Potential competition between retailers and banks for consumer stablecoin balances.
6. Programmable Money: Real-World Examples and Impact (11:58–14:54)
- Stablecoins = Smart, programmable money.
- Payments can be conditional and respond to real-world events (e.g., release money only after package delivery, or automatically optimize bill/rent/savings transfers).
- Loyalty and incentive programs become dynamic:
- E.g., loyalty points could have variable value depending on when/how they are used (13:19).
- Quote:
“They can say, like, your $1 of Dunkin’ Donuts points will be worth $2 if you use it by Friday ... or $10 if you hold on to it for a month to encourage retention.”
— Grace Broadbent (14:21)
7. Debate: Will AI Agents Push Adoption of Stablecoin Rails? (16:20–22:58)
Round 1 – Infrastructure vs. Consumer Behavior
- Grace’s thesis (Stablecoin rail camp):
- AI agents work best with programmable, instant, always-on money (i.e., stablecoins).
- Legacy rails (ACH, cards) are too slow, unavailable off-hours, and poorly suited for micropayments.
- Quote:
“AI works 24/7 and can make decisions in milliseconds. The banking system...takes holidays off, weekends, batch payments...Stablecoins can keep up with the pace of AI.”
— Grace Broadbent (16:20)
- Quote:
- Myra’s thesis (incumbent rails camp):
- Most consumers don't care how money moves as long as outcomes are positive and transparent.
- Changing rails introduces complexity; inertia and satisfaction with the status quo are strong.
- Quote:
“Most consumers are more than happy with the way things are right now.”
— Myra Thomas (17:37)
- Quote:
- Host Rob notes: The “killer use cases” for stablecoins (micropayments) often aren’t feasible on today’s systems (18:12).
Round 2 – Will New Payment Rails Help or Hinder Consumer Comfort with AI?
-
Myra:
- Education is an ongoing challenge; most people have little idea how payments work “behind the scenes.” More complexity may stoke suspicion.
- Quote:
“Do they [consumers] even know what that means? … No, they don't.”
— Myra Thomas (19:39)
-
Grace:
- Stablecoins are “code already” and seamlessly integrate into AI-driven automation — ideal for complex, conditional, and multi-step tasks (e.g., bill pay right after a paycheck arrives, dynamic transfers to savings).
- Quote:
“Stablecoins are essentially code already. ... The logic can be built directly into the transfer.”
— Grace Broadbent (20:49)
-
Rob’s insight:
- For adoption, the tasks that bring clear efficiency are key (“canceling subscriptions, everybody doesn't want to do it”). Complexity must be hidden from the user.
Notable Quotes & Memorable Moments
- On consumer trust and AI:
“Consumers are fickle lot and they're also suspicious and they're always going to wonder if they're going to get the best deal.”
— Myra Thomas (05:30) - On the stepwise approach to AI agent adoption:
“It has to start with those smaller steps and give consumers a real incentive to use it...building in those smaller steps is really a good way to grow trust.”
— Grace Broadbent (07:40) - On programmable money transforming loyalty programs:
“They love the unused points and things like that.”
— Rob Rubin (14:48) - On the risk to banks:
“If the banking system can't offer money that moves at the same speed, AI agents will simply bypass banks entirely.”
— Grace Broadbent (23:12)
Final Takeaways: What Should Banks Do Next? (23:12–24:12)
- Grace:
- Banks must either innovate with AI and programmable money or risk being bypassed by tech and fintech players.
- Myra:
- Banks need to find their place in this new, AI-driven financial ecosystem; adaptation is critical but there’s still time.
- Quote:
“Banks need to figure out their place in this new ecosystem while fintechs and big tech is really sort of circumventing their place in the agentic world.”
— Myra Thomas (23:26)
- Quote:
- “It’s not all bad news for banks,” Grace adds — there’s time to get strategy and consumer experience right.
- Banks need to find their place in this new, AI-driven financial ecosystem; adaptation is critical but there’s still time.
Timestamps for Important Segments
- 01:21 – Delegate or Don’t You Dare (automation icebreaker)
- 03:17 – “Banking Trends to Watch in 2026” (AI adoption stats/discussion)
- 05:18 – Consumer reluctance/trust in AI
- 07:00 – “Use cases first” approach for AI agent adoption
- 09:33 – Stablecoin rails and regulatory change
- 11:58 – Programmable money and loyalty programs
- 16:20 – Debate: Will AI agents naturally adopt stablecoin rails?
- 19:39 – Consumer understanding of payment rails
- 23:12 – Analyst takeaways and advice for banks
Summary
This episode delivers a lively, optimistic-yet-cautious look at how AI agents and stablecoin-enabled, programmable money are transforming banking. The experts agree: user education, transparency, and incremental, high-value use cases are critical for trust and mass adoption. Banks that fail to keep up with the speed of programmable AI money risk being left behind — but it’s not too late for smart, strategic investment and consumer-centric design.
