{"id":5017,"date":"2026-10-06T20:46:21","date_gmt":"2026-10-07T00:46:21","guid":{"rendered":"https:\/\/paulsyng.com\/blog\/?p=5017"},"modified":"2026-10-06T20:46:23","modified_gmt":"2026-10-07T00:46:23","slug":"when-switching-banks-becomes-a-conversation","status":"publish","type":"post","link":"https:\/\/paulsyng.com\/blog\/when-switching-banks-becomes-a-conversation\/","title":{"rendered":"When switching banks becomes a conversation"},"content":{"rendered":"\n

Banks have spent decades making banking easier while leaving the customer responsible for doing it. The branch became a website. The website became an app. You could check your balance at midnight, deposit a cheque from your kitchen and move money without speaking to anyone.<\/p>\n\n\n\n

But you still had to know what to do.<\/p>\n\n\n\n

You had to notice that your savings rate was poor. Understand whether another offer was better. Read the conditions. Work out what moving would affect. Open the account, complete the forms and make sure nothing broke.<\/p>\n\n\n\n

Banking became more convenient. Managing your financial life remained a job. A surprising amount of the industry\u2019s protection sits inside that distinction.<\/p>\n\n\n\n

Imagine someone with a paycheque arriving at one bank, savings at another, a mortgage coming up for renewal and a credit card they chose years ago. They suspect they could arrange things better. They might even know they are paying too much.<\/p>\n\n\n\n

But finding out means giving up an evening. Acting on it might take a weekend. And if they get something wrong, the consequences are theirs. So they leave it until later. The bank keeps the deposits. The customer keeps the homework.<\/p>\n\n\n\n

In \u201cStuck is not loyal,\u201d<\/a> I wrote about why staying tells us so little about preference. Customers can stay because they love a bank, because leaving is difficult or because they have stopped expecting anything better.<\/p>\n\n\n\n

There is another question underneath that argument. What happens when the customer can hand the work to someone else?<\/p>\n\n\n\n

That is the change I was pointing to in \u201cThe End of Doing Your Own Banking.\u201d<\/a> It deserves a closer look, because much of the evidence used to dismiss it comes from a world in which the customer still had to do the work. The familiar objection is that making switching easier does not make people switch.<\/p>\n\n\n\n

There is evidence for that. The United Kingdom has spent years improving account switching<\/a>. Its Current Account Switch Service handles the transfer of payments and balances within a seven-working-day process. In the final quarter of 2025, it completed 99.2% of switches within that timescale. Across the year, it handled about 1.05 million switches, compared with 1.2 million in 2024. Better switching infrastructure did not produce uninterrupted growth.<\/p>\n\n\n\n

But look at what that service solves. Once someone has chosen a new bank and started the process, it helps carry out the move. That is valuable. It removes real administrative work and reduces the risk of payments going astray.<\/p>\n\n\n\n

The customer still has to reach the decision. They have to notice a reason to look. Find an alternative. Understand it. Decide whether the benefit is worth the trouble. Believe the new arrangement will suit their life.<\/p>\n\n\n\n

Keeping your phone number when you change providers solves a similar part of a similar problem. It removes a reason to stay. It does not compare your household\u2019s plans, work out what you need or decide when an offer is worth taking.<\/p>\n\n\n\n

Portability makes an exit usable. Someone still has to decide to use it.<\/p>\n\n\n\n

This matters when we interpret the evidence. Low switching cannot tell us, on its own, how much staying reflects satisfaction, small potential gains, uncertainty or the effort of investigating. And it cannot settle what happens when the effort changes hands.<\/p>\n\n\n\n

Much of the historical evidence describes customers responding to easier switching while remaining responsible for deciding whether to switch. AI makes it possible to change both. And the adoption question changes too.<\/p>\n\n\n\n

Earlier generations of financial technology asked people to become comfortable doing familiar things in unfamiliar places. Enter your card details on a website. Access your account through a browser. Download the bank\u2019s app. Learn where everything lives. Each institution needed customers to learn its way of doing things. <\/h3>\n\n\n\n

An AI assistant can arrive in someone\u2019s financial life through a different door. OpenAI reported in May 2026 that more than 200 million people came to ChatGPT each month with questions about budgeting, investments, financial choices and goals. <\/h3>\n\n\n\n

They may already use it to understand a document, plan a trip, work through a problem or organize their business. Asking about a confusing bank charge becomes another conversation in a place they already know.<\/p>\n\n\n\n

The interface does not need to explain what a menu item means. The person can ask what they actually want to know.<\/p>\n\n\n\n

\u201cCan I afford this?\u201d<\/p>\n\n\n\n

\u201cWhy am I paying so much interest?\u201d<\/p>\n\n\n\n

\u201cIs there a better place for the money I\u2019m keeping for next year?\u201d<\/p>\n\n\n\n

This behaviour is already visible. OpenAI<\/a> reported in May 2026 that more than 200 million people came to ChatGPT each month with questions about budgeting, investments, financial choices and goals. Its personal finance experience lets eligible U.S. users connect financial accounts through Plaid and ask questions using their own financial information. Those connections currently provide information; they do not let ChatGPT change the accounts.<\/p>\n\n\n\n

Perplexity has also expanded its Plaid integration to include bank accounts, credit cards and loans alongside investments. Users can bring information from different institutions into a conversation about their finances.<\/p>\n\n\n\n

Intuit, whose products include QuickBooks, has announced a partnership with OpenAI intended to bring financial insights and actions through Intuit apps into ChatGPT. The direction is already being built into products and partnerships.<\/p>\n\n\n\n

These developments establish something narrower, and more useful, than a claim that everyone is ready to hand over their money.<\/p>\n\n\n\n

People are already asking AI for financial help. Some are willing to bring their financial information into that relationship. Financial services are entering an interface people use for other parts of their lives.<\/strong><\/p>\n\n\n\n

For those users, the learning curve has already begun elsewhere. The industry can debate how far that trust will extend. It cannot sensibly treat the relationship as something that hasn’t started yet.<\/p>\n\n\n\n

Now imagine the next step.<\/p>\n\n\n\n

A customer has connected the relevant accounts, completed the required identity checks and given a financial assistant permission to monitor them. They have explained what matters: keep enough money available for bills, protect the emergency fund and avoid arrangements that need constant attention.<\/p>\n\n\n\n

The assistant can see the accounts the customer has authorized. It knows when the paycheque arrives, which payments recur and when the mortgage renews. It can ask about anything missing.<\/p>\n\n\n\n

One day, it says:<\/p>\n\n\n\n

\u201cYour savings could earn more elsewhere after fees. I\u2019ve checked the withdrawal conditions and the effect on your other accounts. You would still have enough available for upcoming payments. Would you like me to arrange the move?\u201d<\/p>\n\n\n\n

The customer asks a question. The assistant explains the trade-off. Then the customer says yes. With the necessary permissions and account capabilities, that approval could authorize the service to carry out the steps, track progress, and confirm completion.<\/p>\n\n\n\n

There would still be work. <\/p>\n\n\n\n

Institutions would still need to verify identity, open accounts, move funds, and handle exceptions. Some steps would require additional confirmation. But much of that work would no longer be the customer\u2019s project to manage. That difference is something a switching statistic from ten years ago cannot capture.<\/h3>\n\n\n\n

The old question was whether someone would spend their evening researching a better account and organizing the move. The new question could be whether they would approve a suitable change that a service they trust has already investigated and can carry out.<\/p>\n\n\n\n

Those are different decisions. They ask different amounts of the person making them. And the assistant does not need to wait for the customer to become frustrated enough to start looking. With permission, it can notice that a promotional rate has expired, a balance has grown, or a change in income makes the current arrangement less suitable. The search itself becomes something the customer can delegate.<\/p>\n\n\n\n

There is a further step beyond individual approvals. A customer could give the service an ongoing instruction:<\/p>\n\n\n\n

\u201cKeep enough in my everyday account for bills and a buffer. Review the rest for better savings options. Ask me before opening anything new, and leave my mortgage alone until renewal.\u201d<\/p>\n\n\n\n

That instruction expresses a goal and its boundaries.<\/p>\n\n\n\n

The customer would not need to repeatedly translate it into products, institutions and administrative tasks. The service would monitor the arrangement and flag decisions that need attention.<\/p>\n\n\n\n

Of course, people have delegated financial work before. Advisers, brokers and other services already do parts of this job. The opportunity is to make ongoing help available for ordinary decisions across a person\u2019s financial life, including decisions too small to justify a meeting or too tedious to keep revisiting.<\/p>\n\n\n\n

A modest improvement may never feel worth another evening of research. It can still be worth having a trusted service identify and arrange. This is why focusing only on switching misses part of the change. The more consequential development is continuous attention on the customer\u2019s behalf.<\/p>\n\n\n\n

Canada\u2019s regulatory changes matter because they can extend what an authorized service is allowed to do.<\/p>\n\n\n\n

Sharing information lets a service understand an account. Permission to initiate an action lets it help change the arrangement. Budget 2025 committed the government to legislating the next phase of consumer-driven banking, including actions such as account switching and bill payments, by mid-2027, once the Real-Time Rail is live and widely used. That is a commitment to extend the framework, not a promise that every account and financial product will be fully manageable through an AI assistant on that date.<\/p>\n\n\n\n

The timing and scope of implementation will matter. A mortgage will still have terms. A lender will still make credit decisions. A service cannot move an account through a capability that does not exist.<\/p>\n\n\n\n

But those limits do not require us to ignore the change already underway. The relationship can develop while the ability to act is still incomplete.<\/p>\n\n\n\n

A person can begin with an explanation, add financial context, receive useful recommendations and later authorize a specific action when the service can perform it. Regulation can enable the next step in an existing behaviour. It does not have to create that behaviour from scratch.<\/p>\n\n\n\n

Trust will develop in stages.<\/p>\n\n\n\n

TD\u2019s 2026 U.S. survey illustrates the distinction. 55% of respondents said they used AI to help with financial management decisions. Only 18% said they would trust AI to make financial recommendations on its own. Use was growing faster than willingness to give AI independent responsibility.<\/p>\n\n\n\n

That gap matters. It also gives us a more realistic starting point than imagining a sudden leap to complete autonomy. Someone can want help finding and carrying out a better option while retaining the final decision. They can authorize a limited task, see whether it works and decide what to delegate next.<\/p>\n\n\n\n

The service earns more responsibility through experience.<\/p>\n\n\n\n

For that to work, its explanations must be understandable, its calculations dependable, and its authority clear. When something goes wrong, the customer needs someone accountable for fixing it.<\/p>\n\n\n\n

A friendly conversation cannot substitute for those things. Neither can it substitute for aligned incentives. If an assistant earns more by recommending one institution, the customer needs to understand how that affects the advice. A service that claims to search for the best arrangement must explain which providers it considers and how it gets paid.<\/p>\n\n\n\n

The central promise is that it works for the customer across institutions. Its business model has to support that promise.<\/p>\n\n\n\n

If that service succeeds, the relationship with the bank changes.<\/p>\n\n\n\n

Today, customers often organize their finances around institutions. This is my bank. This is my lender. This is my investment platform. Each has an account, an interface and a set of things the customer needs to remember. An independent assistant could organize those institutions around the customer instead.<\/h3>\n\n\n\n

The person explains their circumstances once, updates them when life changes and asks the same service for help. Underneath that relationship, different providers supply the right products.<\/p>\n\n\n\n

The everyday account might remain where it is. Savings might move. A different lender might become suitable at renewal. An existing product might remain the best choice for years.<\/p>\n\n\n\n

The relationship with the assistant could stay continuous through all of it. This is how banks could lose influence without seeing every customer close every account.<\/p>\n\n\n\n

The paycheque might still arrive. The app might still be installed. The customer might still describe the institution as their bank. But the place where they ask questions, weigh choices and decide what happens next could belong to someone else. A retention dashboard would struggle to show the moment that changed.<\/p>\n\n\n\n

That service might come from OpenAI, Anthropic, Perplexity or a specialist financial company. The eventual winner is uncertain.<\/p>\n\n\n\n

The role is easier to describe.<\/p>\n\n\n\n

It would understand the customer\u2019s circumstances, compare suitable options across providers and carry out authorized decisions. It would preserve the customer\u2019s preferences when the underlying products changed.<\/p>\n\n\n\n

A bank can build a capable assistant too. The difficult question is what that assistant is allowed and incentivized to recommend. If another institution offers a better arrangement for the customer, will it help them move? A bank that wants to become the trusted coordinator of someone\u2019s financial life will eventually have to answer that question through its behaviour.<\/p>\n\n\n\n

Meanwhile, banks will still have substantial work to do. They will hold deposits, extend credit, manage risk and resolve problems. Their reliability, terms and service will still matter. A customer may explicitly tell an assistant to keep money with an institution they trust.<\/p>\n\n\n\n

But providers would face a different kind of competition if a customer\u2019s representative could keep reviewing their suitability.<\/p>\n\n\n\n

An expired offer would be noticed. An unnecessary fee would be questioned. A better product would reach the customer without depending on the customer remembering to look. The advantage of being the existing provider would become harder to separate from the value of remaining the right provider.<\/p>\n\n\n\n

The assumption I would question is that people must become more interested in banking for banking to change. They may remain just as busy. Just as uninterested in comparing account conditions. Just as reluctant to spend a Sunday reorganizing their finances.<\/p>\n\n\n\n

Delegation works with that behaviour.<\/p>\n\n\n\n

People can care about the outcome while wanting less responsibility for the process. They can want control over their money without making financial administration a hobby.<\/p>\n\n\n\n

For decades, banks benefited from the distance between knowing there might be a better arrangement and acting on it. Better websites shortened parts of that distance. Apps shortened more. Portability improved the move itself.<\/p>\n\n\n\n

An authorized AI service could take responsibility for much of the journey: noticing, investigating, explaining, arranging and checking that the result matches what the customer approved.<\/p>\n\n\n\n

The customer could stay with the same assistant for years while their financial products changed whenever there was a good reason. That is the relationship banks are at risk of losing. Their next competitor may be the service that finally takes banking off the customer\u2019s to-do list.<\/p>\n","protected":false},"excerpt":{"rendered":"

Banks have spent decades making banking easier while leaving the customer responsible for doing it. The branch became a website. The website became an app. You could check your balance at midnight, deposit a cheque from your kitchen and move money without speaking to anyone. But you still had to know what to do. You […]<\/p>\n","protected":false},"author":1,"featured_media":5018,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_coblocks_attr":"","_coblocks_dimensions":"","_coblocks_responsive_height":"","_coblocks_accordion_ie_support":"","footnotes":"","rank_math_title":"","rank_math_description":"","rank_math_canonical_url":"","rank_math_focus_keyword":""},"categories":[75],"tags":[],"class_list":["post-5017","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-feature"],"_links":{"self":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/posts\/5017","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/comments?post=5017"}],"version-history":[{"count":1,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/posts\/5017\/revisions"}],"predecessor-version":[{"id":5019,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/posts\/5017\/revisions\/5019"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/media\/5018"}],"wp:attachment":[{"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/media?parent=5017"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/categories?post=5017"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/paulsyng.com\/blog\/wp-json\/wp\/v2\/tags?post=5017"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}