← All summaries

AI Is Coming for Bank Deposits

Banking With Interest · Rob Blackwell, Barbara Rehm — Neil Stanley · May 7, 2026 · Original

Most important take away

AI is fundamentally changing the deposit pricing game by giving depositors the math and motivation to arbitrage their banks, especially CDs with low penalties on high rates. Banks must move away from static rate sheets and commodity pricing toward hyper-personalized, dynamic deposit products, and they should reassess CD early withdrawal penalties immediately because the next rising rate environment will expose how irrationally most CDs are structured.

Summary

This episode delivers actionable insights for bankers, investors, and depositors navigating a market where AI is rapidly removing friction from deposit shopping and rate optimization.

Actionable insights for bankers:

  • Engage with AI now, not later. Stanley specifically calls out that relying solely on Microsoft Copilot is insufficient; bankers should actively use generative AI tools such as Claude, ChatGPT, Perplexity, and NotebookLM, and progress toward agentic AI for compliance, underwriting, and process automation.
  • Eliminate static rate sheets. Publishing fixed rates with constant “specials” trains customers to be rate shoppers and accelerates commoditization.
  • Reassess CD early withdrawal penalties immediately. Most penalties are inverted in economic logic: a one-year-of-interest penalty on a 6% CD is huge, while the same structure on a 1% CD is trivial. Banks holding high-rate CDs should consider waiving penalties to let depositors out, since the bank could refund those balances at lower wholesale rates (e.g., FHLB ~3.8%) and improve margin.
  • Adopt enhanced withdrawal options (Stanley holds two US patents in this area) that make CDs behave more like Treasuries, paying bonuses for early withdrawal if rates fall, which decommoditizes the offer.
  • Use hyper-personalization. Build CDs with custom maturity dates, lines of credit on certificates, and tailored structures rather than the standard 6/12/24-month menu.
  • Empower the front line. When a customer cites a competitor rate (e.g., Edward Jones), do not match. Instead, ask when the customer actually needs the money and structure a custom-term product priced off the FHLB curve.

Actionable insights for depositors and investors:

  • Run the math on existing CDs using AI tools. Stanley explicitly notes that ChatGPT, Claude, and Perplexity will calculate whether breaking a CD penalty and reinvesting at higher rates is profitable, including matching the original maturity date.
  • Prefer Treasuries over comparable-yield CDs when given the choice. Stanley was shocked that 80% of bankers in his LinkedIn poll picked a 5-year, 5% CD over an equivalent 5-year, 5% Treasury, ignoring the Treasury’s optionality (sell at gain if rates fall, no inverted-penalty problem).
  • Consider that interest rates likely will not return to the lows of the past 15 years. Stanley argues a high-FOMO, impatient society correlates with higher, not lower, rates. The 30-year Treasury is near 5% and the yield curve has un-inverted. At the recent FOMC, four dissenters wanted to drop “neutral” language and signal upward bias, surprising market consensus.

Investments and companies mentioned:

  • US Treasuries (preferred over CDs for optionality)
  • Stablecoins under the Genius Act: framed as a defensive concern, not an opportunity, for community banks. Stablecoin issuers invest 100% of reserves in Treasuries vs. banks’ fractional Treasury holdings, redirecting deposit flows.
  • Visa and Mastercard: stablecoin payment rails threaten interchange revenue.
  • Fintechs cited as deposit competition: PayPal, Venmo, Chime.
  • Intuit, TurboTax, QuickBooks, Credit Karma plus OpenAI integrations: a likely future channel for AI-driven deposit movement recommendations.
  • AI tools recommended for bankers: Claude, ChatGPT, Perplexity, NotebookLM, HeyPex, Quad (cowork agents).
  • Bank of North Dakota: cited as one of the few banks engaging proactively with stablecoin defense.

Why these matter: the combination of AI-driven optimization, stablecoin payment rails, and a higher-for-longer rate environment means deposit franchises that look stable today are far more vulnerable than balance sheets suggest. Banks that fail to redesign products and price dynamically will lose the highest-rate, highest-cost deposits first while keeping the worst ones, the opposite of what their CD penalty structures were designed to do.

Legislative watch: The Tillis-Alsobrooks compromise language on stablecoin yield is being challenged by ABA, BPI, CBA, and ICBA as having “massive holes.” A Senate Banking Committee vote on the crypto market structure bill is uncertain because Senator John Kennedy is conditioning his vote on House action on a separate housing bill. Investors in bank stocks should monitor this closely; passage in current form would accelerate deposit outflows to stablecoin.

Chapter Summaries

  1. Introduction and guest setup. Rob Blackwell hands the mic to Barbara Rehm to interview Neil Stanley, founder of The CorePoint, on how AI, sophisticated customers, and CD repricing are reshaping deposit competition.

  2. Why bankers cannot wait on AI. AI is the next-generation toolset and behaves more like a workforce than installed software. Non-regulated industries are moving faster, and banks risk irrelevance if products and processes look the same in five years.

  3. AI’s impact on deposit pricing. Generative AI is non-deterministic and pattern-based. Combined with open banking and overnight sweeps, friction has been removed and depositors are actively reconsidering where money sits, especially younger generations who do not default to bank deposits.

  4. Hyper-personalization and dynamic pricing. CDs are seen as boring by 35-year-olds. Banks must build for individual use cases, replace static monthly rate sheets with dynamic systems, and equip both bankers and customers with better analysis tools.

  5. Leadership, trust, and the regulator angle. AI adoption requires leadership from the very top. The winners combine tech efficiency with a concierge-like personal trust model. Bankers can no longer rely on information asymmetry because customers now have powerful knowledge bases too.

  6. Why CDs are uniquely vulnerable. CDs are about 20% of deposits. Most early withdrawal penalties are economically inverted, creating arbitrage opportunities. AI tools will run the numbers, identify profitable break-and-reinvest moves, and eventually execute the moves through agentic integrations with platforms like Intuit and OpenAI.

  7. Defensive strategies. Banks should hyper-personalize, decommoditize, and adopt enhanced withdrawal options that pay bonuses if rates fall. Stanley’s LinkedIn poll showing 80% of bankers preferring a 5% CD to a 5% Treasury illustrates how poorly the industry understands optionality.

  8. Interest rate outlook. Stanley argues the impatient, FOMO-driven society correlates with higher rates. The yield curve has shifted, the 30-year is near 5%, and four FOMC dissenters wanted hawkish language, contradicting consensus that rates are heading down.

  9. Generative vs. agentic AI. Generative AI is the thought partner; agentic AI refines it for controlled, deterministic-feeling processes such as loan underwriting or compliance testing. Dynamic deposit pricing can be built without full agentic AI today.

  10. The Genius Act and stablecoin. Not designed to benefit banks. Stablecoin funds the US Treasury and creates a new payment rail that bypasses card interchange. Banks should focus on defensive moves and lobby through their associations.

  11. Practical dynamic pricing example. When a customer cites a competitor’s rate, do not match. Ask when they actually need the money and build a custom-term product priced off FHLB rates, decommoditizing the conversation.

  12. Wrap-up priorities for bankers. Engage with multiple AI tools beyond Copilot. Kill the static rate sheet. Get the C-suite up to speed on stablecoin. Train the front line.

  13. Fair lending and compliance. Banks already do not use one-rate-fits-all on deposits. Fair lending protects vulnerable consumers from deceptive practices, not negotiation outcomes. Paying differing rates based on negotiation is not disparate impact.

  14. Closing news segment. Rob Blackwell covers the Tillis-Alsobrooks stablecoin yield compromise, banking trade group opposition, and uncertainty around a Senate Banking Committee vote on the crypto market structure bill.