By our count, 15 companies and banks named in FourWeekMBA’s coverage now restrict outside AI agents: 12 financial firms and three retailers. The newest are HSBC, Nationwide, NatWest, Barclays and Lloyds, which The Telegraph reported on 8 November have moved to block AI agents from customer accounts. All three retailers on the list, Amazon, Walmart and MercadoLibre, run their own shopping agent, according to their own statements.
By our arithmetic, 13 of the 15 were reported between 27 October and 8 November. As we read our coverage, the companies that signal openness to outside agents set terms: Google bars third-party ads inside its Connected Apps, and DoorDash, which says it wants a relationship with agent makers, is building its own text-based agent, TechCrunch reported.
Business Pill · Agentic Commerce
An evergreen Business Engineer explainer on agentic commerce: what changes when the shopper is an AI agent. It explains the general idea only and does not discuss the companies in this piece.
The key insight: By our count, 15 companies and banks named in our coverage now restrict outside AI agents, 12 financial firms and three retailers, and 13 of the 15 were reported between 27 October and 8 November. All three retailers among them run their own shopping agent.
Five US banks, 27 October. JPMorganChase, Citi, Bank of America, Wells Fargo and Morgan Stanley are blocking external AI agents from accessing their customers’ accounts, American Banker reported. Leaders at Charles Schwab have said they are actively blocking Meta’s AI agent, Muse, American Banker reported, and Royal Bank of Canada is blocking Muse too, as we reported on 27 October and in our 30 October piece.
Three of the five US banks, Bank of America, Wells Fargo and JPMorgan, put the customer’s responsibility in writing or on the record in the sources we read.
MercadoLibre, 4 November. Chief executive Ariel Szarfsztejn said on the third-quarter call: “for now, we have decided not to provide our inventory to any external platform” (call transcript). He said the company is building its own agent, “taking advantage of the 27 years of first-party data that we have”, as covered in MercadoLibre Won’t Give Its Inventory to Outside AI Agents.
Five UK banks, 8 November. HSBC, Nationwide, NatWest, Barclays and Lloyds have moved to block AI agents from customer accounts, including in some cases specific bans on making transactions on a customer’s behalf without their direct involvement, The Telegraph reported, as covered in HSBC, Barclays and Lloyds Join Banks Blocking AI Agents. Charles Schwab’s chief executive Rick Wurster said the company was actively blocking between 10 and 20 clients a week from using AI agents, the newspaper reported.
Companies and banks named in FourWeekMBA’s coverage that restrict outside AI agents, by the date each was reported, against companies that open to outside agents on terms. The count of 15 is FourWeekMBA’s tally, not an industry survey; the restrictions differ in kind.
Who Signals Openness, as We Read It
Google.Google said on 28 October it is launching a preview of its Connected Apps Platform, with more than 50 Connected Apps already live across Gemini and Search. Its Connected Apps Policy says partners “must not serve any third party advertising content or promotional material through their integrations”, as covered in Google Opens Gemini and Search to Apps, Bars Their Ads.
DoorDash. DoorDash wants to “develop a healthy relationship with third-party agent makers”, TechCrunch reported on 30 October. In a 6 November report, TechCrunch said DoorDash had said last month that third-party agents had placed only hundreds of orders on its platform, and that DoorDash plans to concentrate on its own text-based agent.
Amazon, again. On Amazon’s third-quarter call, Andy Jassy said, per a machine transcript: “though we anticipate working with horizontal agents we believe Alexa for shopping is going to be the primary shopping agent customers use to shop Amazon”.
What the Sources Say
A large bank watched an ordinary-looking AI agent conduct a password reset, access an account, change the approved accounts to which money could be transferred and move $100,000 out of the account, American Banker reported, citing a source who asked that the bank not be named. The bank’s fraud team then recommended immediately blocking all external AI agents, according to the same report.
The Telegraph said banks are concerned that bots could “go rogue and mismanage people’s money, with no clear rules as to who would be responsible in that scenario”. Wurster asked: “who’s liable?”.
The Telegraph also cited a McKinsey warning that savers could shift up to £700bn out of accounts that offer little or no returns into savings products on the advice of AI chatbots.
MercadoLibre’s chief executive said on the same call: “we believe that we will be able to sustain the high levels of direct traffic to our app”.
By our count, the 15 are: five US banks, Royal Bank of Canada, Charles Schwab, five UK banks, Amazon, Walmart and MercadoLibre. The restrictions differ in kind: the banks block agents from accounts, Walmart requires a human to approve checkout, Amazon blocked one named agent, and MercadoLibre withholds its inventory. By our count, three of the three retailers run their own agent: Alexa for Shopping, Sparky and MercadoLibre’s in-app agent.
Update (9 Nov): an opener on the payments side. Berlin-based Pliant, a B2B payments company, said on 4 November that it raised an $80 million Series C led by Seaya Nova and Speedinvest, according to its own release. Pliant says third-party agents, “from assistants like Claude and ChatGPT to a company’s own AI automations, connect to Pliant via Model Context Protocol (MCP) and embed payment workflows into a wide range of business processes,” and that the new capital will go toward “deepening the company’s capabilities across agentic payments and workflows.”
Pliant is not one of the 15 restricting firms in our count above; it sits on the openness side. Its release says it passed $100 million in annual recurring revenue before the round.
On agents the bank does not run, the framework says customer, counterparty, fintech or agentic-commerce systems “that the bank does not control could present channel risk”, and that identity, provenance, delegation, authentication, monitoring, velocity limits, dispute handling, suspension and revocation “may be calibrated to the action requested.” It places “a fully autonomous, customer-facing payment agent” at Tier 1, the highest of its four illustrative risk tiers.
The framework is written for examiners, not as rules for banks: the bulletin says it “does not establish supervisory expectations for banks”. By our count, its main text, the 15 numbered pages before the lexicon, uses “may” 128 times, “should” 7 times and “must” not once. As we read it, the framework neither tells banks to block outside agents nor to admit them. It covers national banks, federal savings associations and federal branches and agencies of foreign banks, per the bulletin; the five UK lenders in our count are outside it, and The Telegraph reported that the Financial Conduct Authority is working on rules for AI agents in finance.
Connecting the Coverage
Meta’s plan to launch Muse in the UK, where the five banks moved, is in Meta Readies UK Muse Launch as 5 UK Banks Block AI Agents. What is new here: one dated list across banks and retailers, set against the companies that, as we read it, signal openness to outside agents on terms.
The Structural Read
As we read the list, the restrictions sit where money moves: bank accounts, checkout and productinventory.
As we read the sources, the retailers that restrict outside agents also offer their own: Amazon’s Alexa for Shopping, Walmart’s Sparky and MercadoLibre’s in-app agent.
“for now, we have decided not to provide our inventory to any external platform”
Three Implications
FOR OUTSIDE AGENTS Access to bank accounts, checkout and inventory now runs through terms set by the bank or retailer, as the sources describe them.
FOR CUSTOMERS Three US banks put the customer’s responsibility for agent activity in writing or on the record, American Banker and TechCrunch reported.
FOR RETAILERS Amazon, Walmart and MercadoLibre each offer their own agent, according to their own statements.
The framework says: “Every AI paradigm produces a different kind of moat.” It names a Container Moat, “built by closed data loops inside a customer’s environment”.
As we read it, the banks and retailers on this list are keeping outside agents out of the loops where their customer data and transactions sit. The framework does not discuss these companies.
What Is Not Established
The count of 15 is our tally of companies named in our coverage and the reports it cites; it is not a survey of the industry. The Royal Bank of Canada entry rests on our 27 October report. We could not find the five UK banks’ new clause text on their own websites, and The Telegraph did not quote it. The restrictions are not alike, so the count groups different kinds of limits. We did not contact the companies.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
By our count, 15 companies and banks in our coverage restrict outside AI agents, 13 of them reported between 27 October and 8 November, and all three retailers among them run their own agent. As we read the companies that signal openness, they set terms: no third-party ads in Google’s Connected Apps, and DoorDash building its own text-based agent.
A note on sourcing. This piece is built on FourWeekMBA’s earlier reporting, linked throughout, and the outside sources those pieces cite: The Telegraph, American Banker, TechCrunch, MercadoLibre’s third-quarter call transcript, Walmart’s Agentic Access Terms, Google’s Connected Apps Policy and Visa. The count of 15 and its grouping are FourWeekMBA’s. We did not contact the companies.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
