Posted on

Banks are giving AI agents more work while keeping a close eye on how far they can go

A recent survey found that 52% of financial-services respondents are actively adopting agentic AI. However, the bigger question now is how much of the actual work these agents are now being trusted to do – moving beyond analysis and recommendations to execution.

This boundary is increasingly central to how these systems are evaluated through model-risk and vendor-review processes, whether banks add agents to banking workflows through vendors like nCino or build and deploy agents on their infrastructure through Google Cloud.

nCino customers see agents as an extension of their workforce

nCino is introducing agents into existing banking jobs rather than treating them as a separate AI layer. nCino’s Digital Partners, a suite of AI agents designed for specific banking functions, includes agents for analyst, processor, and service roles, and the company describes them as a “dual workforce” in which AI handles data-heavy work. At the same time, bankers retain time for judgment, relationships, and growth.

From the bank’s perspective, the important question is therefore which part of a banker’s workflow the institution is comfortable handing over to the agent. A commercial banker, for example, can have an agent collect and analyze information needed for a relationship review and prepare the work, rather than spending hours doing the underlying research manually. The banker then reviews that work and remains responsible for the consequential judgment.

For nCino’s latest customer, ConnectOne Bank, the interest in the vendor’s agentic technology is to free its bankers from administrative work so they can spend more time on higher-value work. 

…

Posted on

NVIDIA bought Hugging Face. What happens to banks when AI models become open?

    Weekly 10-Q

    The weekly 10-Q newsletter is part of the Tearsheet Pro subscription, where I unpack the recent moves and strategies of leading banks and fintechs in the public space, coupled with stock market analysis. In your inbox every Friday!

    Message Sara

    NVIDIA bought Hugging Face. What happens to banks when AI models become open?

    What does this acquisition change for banks?

    A bank deciding which AI model to use is becoming a little like a bank deciding which payment rail to use. The next natural question in line is who controls the infrastructure underneath it. This is what makes NVIDIA’s $12.93 billion acquisition of Hugging Face worth watching from a financial services perspective. What looks like a bet on open-source AI also forces a harder question for banks: does making AI models more open give financial institutions more control, or does it simply move dependence to another layer of the stack?

    There are credible arguments for both scenarios.

    Hugging Face has become a major distribution point for open AI, with more than 18 million developers, 3 million models, 500,000 datasets, and 1 million applications on its platform. More than 200,000 companies use it to discover, evaluate, customize, and deploy AI. NVIDIA says the platform will remain open and that customers will be free to choose their models, frameworks, clouds, inference providers, and computing platforms.

    Financial institutions are already moving in this direction. NVIDIA’s 2026 financial-services survey found that 84% of respondents consider open-source models and software important to their AI strategy, while 42% are using or assessing agentic AI.

    We take a look at two potential scenarios that could unfold for banks following this acquisition.

    Scenario 1: Open models give banks more control

    …

    subscription wall for TS Pro

    Posted on

    Fifth Third is trying to make a bigger bank feel simpler

      Weekly 10-Q

      The weekly 10-Q newsletter is part of the Tearsheet Pro subscription, where I unpack the recent moves and strategies of leading banks and fintechs in the public space, coupled with stock market analysis. In your inbox every Friday!

      Message Sara

      Fifth Third is trying to make a bigger bank feel simpler

      Fifth Third is building more ways to serve customers without adding complexity.

      Fifth Third [FITB] has launched Truly Simple, a new card for easing the cost of carrying revolving debt. The card offers 0% APR on purchases and balance transfers for 18 months, carries no annual fee, and lets customers move balances and manage payments through Fifth Third’s mobile app. 

      “People are looking for practical and reliable ways to manage debt,” Ben Hoffman, Chief Strategy Officer and Head of Consumer Products, said during the announcement of the card launch.

      With its recent moves across AI, small-business banking, embedded finance, and acquiring Comerica, the new card points to a larger ambition for Fifth Third, expanding its reach while making the banking experience feel simpler. 

      The Ohio-based bank is simplifying its consumer card lineup around two primary needs: its 1.67% Cash/Back card for everyday rewards and Truly Simple for customers who want to finance purchases or consolidate debt. The firm is simplifying its product portfolio so customers have a clearer reason to choose one product over another.

      We look at how the new card launch is a part of Fifth Third’s broader product strategy.

      A bigger bank, with more ways into the relationship

      …

      subscription wall for TS Pro

      Posted on

      The Week in Market Moves | Sept 03 – 10, 2026

      Company signals and market response

      This analysis tracks the top company developments and how markets absorbed them through Thursday’s close, focusing on where shifting narratives translate into price action.

      It is part of Tearsheet PRO’s weekly 10-Q Newsletter, where strategy meets market reaction. I track how leading banks and fintechs are evolving in public markets and how investors are pricing those moves.

      Subscribe to PRO and get the full 10-Q story in your inbox every Friday!

      1. Chime (CHYM) – Close: $32.67

      • Chime is paying $590 million in cash to acquire Stride Bank, its bank partner of more than seven years, with the deal expected to close in the first half of 2027, subject to regulatory approval.
      • Stride’s national bank charter will become Chime’s own banking infrastructure, with the bank expected to operate as Chime Bank, N.A., a wholly owned subsidiary.

      Why it matters: Chime is bringing a critical piece of its banking stack in-house. The move reduces its dependence on a partner bank, eliminates partner-bank fees, and gives it more control over how quickly it develops and launches products. More importantly, it changes the economics and structure of the fintech-bank model: Chime is no longer just building on top of a bank; it is buying the bank underneath it.

      2. U.S. Bank (USB) – Close: $62.41

      • U.S. Bank launched USBDC, its own dollar-backed stablecoin, and used it in a live cross-border payment between its North American and European entities.
      • The pilot ran on the Stellar blockchain and tested the full stablecoin lifecycle, including minting, payments, redemption, freezing, and clawback, while connecting back into the bank’s existing risk, compliance, and operating infrastructure.

      Why it matters: The interesting part is that U.S. Bank is testing whether tokenized money can actually run inside a regulated bank’s existing machinery. That makes the experiment less about crypto and more about modernizing cross-border money movement. As more banks issue their own digital dollars, the harder problem may become making those different forms of money interoperable.

      3. Visa (V) – Close: $367.21, Mastercard (MA) – Close: $565.37

      • Visa, Mastercard, and Ant International are working toward a common Know Your Agent framework to help identify and onboard AI agents across payment networks, wallets, marketplaces, and agent platforms.
      • The three companies already have separate agent protocols – Visa’s Trusted Agent Protocol, Mastercard’s Verifiable Intent, and Ant International’s Agentic Mobile Protocol – and are now exploring common principles and trust signals.

      Why it matters: Agentic commerce creates a new identity problem: the party initiating a transaction may no longer be the human customer. If every network builds its own way to verify an agent, merchants and platforms could end up repeating the same checks across ecosystems. A common KYA layer could become the plumbing that lets an AI agent carry trusted identity and risk signals from one payment environment to another.

      4. Coinbase (COIN) – Close: $172.28

      • Coinbase is partnering with Moov to bring stablecoin payments, settlement, and real-time funding to community banks and credit unions.
      • Moov will embed Coinbase’s regulated digital-asset infrastructure into its existing payments platform, allowing institutions to offer stablecoin services without building a separate crypto stack.

      Why it matters: Coinbase is attacking one of the biggest barriers to stablecoin adoption among smaller financial institutions: the technology burden. Instead of asking a community bank to become a crypto infrastructure company, the partnership puts those capabilities behind the payments systems it already uses. That could make stablecoins less of a specialized digital-asset product and more of an infrastructure feature that banks can switch on.

      4. Nu (NU) – Close: $15.02

      • Nu has entered the U.S. with a full consumer banking suite through partner bank Lead Bank, including a 3.5% APY account, credit card, debit card, savings tools, and cross-border transfers.
      • The launch gives Nu a way to enter the market while it works toward its own U.S. national bank charter, for which it received conditional OCC approval in January 2026.

      Why it matters: Nu is taking a different route into the U.S., rather than waiting for its own charter before launching. Lead Bank gives it the regulatory and banking infrastructure to start learning from U.S. customers now, while Nu builds toward owning more of that stack itself. Nu Global, a multicurrency account that uses USDC and EURC to move money across 35+ countries, also gives the U.S. launch a distinctly cross-border angle, extending Nu’s Latin American cross-border DNA into its U.S. strategy.

      Posted on

      Venture banking 3.0: How Stifel is rebuilding trust after Silicon Valley Bank

      When Silicon Valley Bank collapsed in March 2023, it took with it a bank that had quietly become infrastructure for the startup ecosystem, holding the majority of the venture debt market and serving as the one relationship founders and fund managers trusted for everything from a wire transfer to an introduction. Bigger banks and a wave of neobanks moved fast to fill the void. Three years later, the question is whether either one actually replaced what SVB was.

      Today I’m joined by Katya Kohen, managing director at Stifel, where she leads investor coverage nationally for the firm’s venture and fund banking group. Katya’s path into banking runs through venture, not around it. She built and exited a tech company, invested as an angel in immigrant founders, built her own accelerator, and spent years at Techstars before joining Stifel as, in her words, its first non-traditional banker. That mix of operator, investor, and now banker shapes what she calls venture banking 3.0, a founder-first model built after the SVB meltdown.

      We get into what actually broke in the founder-bank relationship, how Stifel has grown its venture banking book toward $12 billion in commitments, and why venture debt has become essential capital for AI companies given how capital-intensive the category has become.

      Let’s get into it.

      Watch the episode

      Listen to the full podcast

      Subscribe: Apple Podcasts I SoundCloud I Spotify

      Top-line Takeaway: Katya Kohen, managing director at Stifel, where she leads the firm’s venture and fund banking group, argues that venture banking is entering a new era after Silicon Valley Bank’s collapse. She explains how Stifel is building what she calls “venture banking 3.0”: combining the trusted human relationships founders still need with the digital experience they now expect, while connecting them to capital, treasury, wealth management, and investment banking as their companies grow. Kohen also makes the case for venture debt as an increasingly important tool for capital-intensive AI companies, and says the SVB failure has made founders more deliberate about who they trust with their money. For Stifel, the opportunity is to bring those pieces together into a founder-first model that supports companies from their earliest funding rounds through an eventual exit.

      Read the whole transcript (for TS Pro subscribers)

      Meet Katya Kohen

      …

      Posted on

      The Quarterly Review: BNY’s Carl Slabicki is pushing toward practical innovation

       

      In this edition, we spotlight Carl Slabicki, Head of Commercial for Global Payments & Trade at BNY. 

      At BNY, Carl Slabicki oversees the firm’s work connecting payments, liquidity, and trade for clients navigating an increasingly always-on financial landscape. BNY has long been a central player in global custody and treasury services, and under Slabicki’s tutelage over these past few months, its footprint has been expanding into more integrated, cross-capability solutions spanning digital assets, data, and AI. 

      In this edition, Slabicki dives into how BNY is moving clients from broad modernization conversations toward practical and measurable execution, and how this shift is shaping the firm’s approach to client relationships, market education, and commercial strategy.

      …

      Posted on

      The Week in Market Moves | Aug 27- Sep 03, 2026

      Company signals and market response

      This analysis tracks the top company developments and how markets absorbed them through Thursday’s close, focusing on where shifting narratives translate into price action.

      It is part of Tearsheet PRO’s weekly 10-Q Newsletter, where strategy meets market reaction. I track how leading banks and fintechs are evolving in public markets and how investors are pricing those moves.

      Subscribe to PRO and get the full 10-Q story in your inbox every Friday!

      1. Affirm (AFRM) – Close: $74.30

      • Affirm and Shopify are bringing Shop Pay Installments to Australia, giving eligible shoppers the option to split purchases into fortnightly or monthly payments.
      • The launch expands Affirm’s Shopify footprint beyond the U.S., Canada, and the U.K. and marks Affirm’s return to Australia, with the service powered exclusively by Affirm.

      Why it matters: Affirm is extending a distribution model that puts its credit product directly inside Shopify’s checkout. That’s a step forward as BNPL moves toward more frequent, everyday transactions rather than being reserved for large-ticket purchases. Affirm’s latest quarter supports that shift, with GMV up 36% to $14.1 billion and transactions growing by 41% to 53 million. The Australia launch gives that broader usage strategy another market to test.

      2. NVIDIA (NVDA) – Close: $228.45

      • NVIDIA has paused some revenue-sharing deals under its AI Compute Partnership Program, less than two months after launching the financing initiative for smaller AI cloud providers.
      • The retreat comes amid concerns that NVIDIA’s role as chip supplier, financier, and potential revenue participant could attract greater antitrust scrutiny.

      Why it matters: NVIDIA’s ambition has been expanding beyond selling chips into helping finance the infrastructure needed to deploy them. That can accelerate the AI buildout, but it also puts the firm in an increasingly unusual position: supplier, capital provider, and economic participant in the businesses using its hardware. The pause suggests there are limits to how far NVIDIA can extend that influence without raising questions about competition and the financing structures supporting AI demand.

      3. Intuit (INTU) – Close: $344.30

      • Intuit is bringing QuickBooks and Mailchimp into Perplexity Computer through MCP connectors, allowing businesses to access Intuit’s data and capabilities from an AI interface.
      • The integration moves beyond retrieving information, with users able to manage cash flow, chase overdue invoices, access payroll information, and analyze marketing performance.

      Why it matters: If businesses increasingly ask AI agents to do the work, Intuit’s value has to travel with the customer rather than remain inside QuickBooks or Mailchimp. The partnership with Perplexity lets Intuit put its financial data and domain expertise into an agentic workflow while keeping its systems underneath the actions being taken. That could make Intuit’s financial intelligence more valuable even when the customer is no longer starting with an Intuit product.

      4. Fifth Third Bank (FITB) – Close: $54.97

      • Fifth Third launched the Truly Simple Credit Card with 0% APR on purchases and balance transfers for 18 months and no annual fee.
      • The bank is simplifying its card lineup around two primary jobs: its 1.67% Cash/Back card for rewards and Truly Simple for financing purchases and consolidating higher-rate debt.

      Why it matters: Fifth Third is assigning each card a clear customer job instead of competing with a crowded rewards-card market on features. That fits the bank’s broader push to make its products feel simpler even as the institution itself becomes more complex through digital expansion and the Comerica integration. The card is a relatively straightforward product move, but the sharper strategy is deciding what each product is actually for.

      Posted on

      Goldman Sachs built its talent pipeline around apprenticeship. AI is now testing what that means

        Weekly 10-Q

        The weekly 10-Q newsletter is part of the Tearsheet Pro subscription, where I unpack the recent moves and strategies of leading banks and fintechs in the public space, coupled with stock market analysis. In your inbox every Friday!

        Message Sara

        Goldman Sachs built its talent pipeline around apprenticeship. AI is now testing what that means

        How do you develop talent when AI does the work they once learned from?

        Goldman Sachs approaches the first months of a young employee’s career as a formative period. The company puts new hires close enough to experienced bankers and traders to learn what no classroom could teach: how a client conversation actually works, why a senior colleague makes a particular call, and what to notice before anyone explains it.

        AI is changing the economics of that first pass and potentially the learning that came with it. The pressure is now showing up in a new place. Goldman’s internal research says entry-level workers are already facing stronger AI-related hiring headwinds than more senior employees. Entry-level work has traditionally served two purposes: getting the work done and training the person doing it. If AI eliminates enough of the former, companies have to deliberately recreate the latter.

        At the same time, Chris Churchman, who heads Marquee, Goldman’s digital platform for institutional clients, warns that the bank could automate away some of the very experiences through which junior employees learn to become senior ones. 

        So, what happens to the apprenticeship model when AI starts doing the work apprentices used to learn from?

        Goldman’s early-career programs were always deliberately hands-on

        When I spoke with Omer Tanvir, Goldman Sachs’ former global head of campus and diversity recruiting, for Tearsheet’s coverage of the bank’s 2024 summer internship program, the word “apprenticeship” described how the firm expected young people to learn.

        …

        subscription wall for TS Pro

        Posted on

        The Week in Market Moves | Aug 20-27, 2026

        Company signals and market response

        This analysis tracks the top company developments and how markets absorbed them through Thursday’s close, focusing on where shifting narratives translate into price action.

        It is part of Tearsheet PRO’s weekly 10-Q Newsletter, where strategy meets market reaction. I track how leading banks and fintechs are evolving in public markets and how investors are pricing those moves.

        Subscribe to PRO and get the full 10-Q story in your inbox every Friday!

        1. Wells Fargo (WFC) – Close: $84.97

        • Wells Fargo is stepping up its recruitment of independent advisers, who can use the bank’s infrastructure without becoming full-time employees; those advisers have already brought in $17 billion this year.
        • The strategy is also helping offset adviser departures across the industry, with Wells Fargo attracting teams such as James Taylor’s from Morgan Stanley, along with nearly $6 billion in client assets.

        Why it matters: The wealth-management model is changing as technology makes it easier for advisers to operate independently. Rather than fight that shift, Wells Fargo is trying to become the infrastructure layer that independent advisers can build on. That gives the bank a way to keep the economics and relationships of wealth management without insisting every adviser fit the traditional employee model. The bigger bet is that flexibility, rather than employment status, becomes the new battleground for adviser talent.

        2. Visa (V) – Close: $379.66

        • Visa has expanded its Visa Vulnerability Agentic Harness (VVAH) from finding and assessing vulnerabilities to actually remediating and validating them.
        • It is pairing the technology with an expanded cybersecurity advisory practice, helping clients assess risk, prioritize vulnerabilities, and build remediation roadmaps.

        Why it matters: The interesting shift here is that AI is moving further down the operational chain. Finding a vulnerability is useful, but the real value comes from shortening the distance between discovery and fixing it. That’s becoming more important as attackers can exploit newly discovered weaknesses in hours rather than weeks. Visa is treating cybersecurity as an ongoing response loop, not a periodic assessment exercise.

        3. TD Bank (TD) – Close: $121.09

        • TD generated C$195 million ($141 million) in AI value during the first three quarters of fiscal 2026, putting it within distance of its full-year C$200 million target months early.
        • The bank is deploying AI across credit, software development, and contact centers, while expanding into areas such as insurance claims, employee knowledge management, and income verification.

        Why it matters: TD’s significance is that AI is being measured as a business outcome rather than an innovation program. The bank is pushing AI into processes where the payoff can show up in lower unit costs, faster decisions, and less manual work. It also gives TD a concrete baseline against which future AI spending can be judged. The next question is whether these early gains can compound to achieve the bank’s C$1 billion medium-term AI value ambition.

        4. Affirm (AFRM) – Close: $77.49

        • Affirm’s transactions grew 41%, faster than its 36% GMV growth, while average order value fell 4% – a sign that consumers are using BNPL for more frequent, smaller purchases.
        • Affirm Card is accelerating that shift: active cardholders more than doubled to 5.2 million, while card GMV jumped 124% to $2.8 billion.

        Why it matters: Smaller baskets and higher transaction frequency suggest Affirm is expanding from a financing product into a broader payment habit. That creates more opportunities for usage, but also puts more pressure on underwriting discipline as frequency rises. The usual test is whether Affirm can increase everyday engagement without turning that broader reach into a credit-quality problem.

        5. NVIDIA (NVDA) – Close: $227.98

        • Nvidia is reportedly nearing a roughly $13 billion acquisition of Hugging Face, the platform where developers share, discover, and build AI models.
        • The deal would give Nvidia a much deeper position in open-source AI at a time when developers are looking for alternatives to models controlled by OpenAI, Anthropic, and other closed-model providers.

        Why it matters: This is a strategic move beyond chips. Nvidia’s hardware dominance ultimately depends on there being a large and growing ecosystem of models that need to run on that hardware, and open-source models can help expand that market. Bringing Hugging Face closer could give Nvidia influence over both the developer and compute layers. It also highlights how the AI infrastructure battle is broadening: firms are now eyeing controlling the ecosystem around how models are built, deployed, and run.

        Posted on

        Live Oak Bank’s BJ Losch on why AI is an accelerant, not a strategy

        Most banks seem to chase small business customers as one segment among many. Live Oak Bank was built around a bet that specializing beats generalizing — and it started about as narrow as a bank can start, lending only to veterinarians. Today, Live Oak has grown that thesis into 40 verticals, holds the top spot among all SBA 7(a) lenders by dollar volume, and has done it all without a single branch.

        My guest is BJ Losch, president of Live Oak Bank. Live Oak has never had a physical location, yet its people travel the country to sit across the table from customers a branch down the street never would. Now the bank is layering AI onto that model, cutting loan approval-to-close timelines from over 100 days toward a two-week target, and thinking hard about what it means to stay high-touch while going high-tech.

        We talk about the theory of verticality, where automation should and shouldn’t touch the credit decision, and why BJ sees AI as an accelerant of Live Oak’s strategy rather than the strategy itself.

        Watch the episode

        Listen to the full podcast

        Subscribe: Apple Podcasts I SoundCloud I Spotify

        Top-line Takeaway: In his conversation with Tearsheet founding editor Zack Miller, BJ Losch, President of Live Oak Bank, underscores that AI works best as an accelerant to a strong strategy, not a strategy in itself. For Live Oak, the branchless, nationwide bank, that strategy is vertical specialization: deeply understanding small businesses in specific industries and pairing that expertise with high-touch service. AI can take on more of the manual work behind lending, helping the bank move toward two-week loan closings and making personalized service more scalable. But Losch draws a clear line around human judgment in credit decisions, where trust, accountability, and borrower context still matter. The bigger opportunity, he says, is using AI to rethink workflows, productivity, and distribution while preserving the customer intimacy that has differentiated Live Oak from traditional banks.

        Read the whole transcript (for TS Pro subscribers)

        The backstory behind becoming the top SBA lender

        …