A community bank on the US East Coast runs an AI customer service chatbot on its public website, grounded entirely in the bank’s own published rates, terms and product material. It has answered 14,586 questions to date, resolved 91.6% of them from that material over the last six months, and answered at least one question on each of the last 330 consecutive days.
Its digital marketing team owns it outright, and in two years of operation the bank has raised zero support tickets. The bank shared the deployment with us anonymously.
The questions keep arriving after the branch has shut
A customer at eleven at night is working out whether a certificate renews on its own or whether she has to do something. A small-business owner is comparing two business checking accounts on a Sunday, deciding which one to open in the morning. Somebody cannot get into online banking and does not want to wait until Monday to ask why.
The bank has the answer to all three. It is published, it is approved, and it is sitting on the bank’s own website.
The problem is that the website was built to present products rather than to answer questions, so finding the answer means already knowing which page it lives on. The bank’s marketing team could see the gap clearly enough.
The questions were arriving, and the site was replying with a navigation menu.
24/7 customer self-service, restricted to published material
One assistant sits in live chat on the public site, open to anyone, and effectively every question arrives through that single channel. It answers from the bank’s rates, product terms, account types, branch information and online banking help, and from nothing beyond them.
There is no core banking connection. No custom build, no integration program, no standing internal headcount against it. The digital marketing team runs it, and has hand-written 58 revisions to the assistant’s persona since launch.
That ownership model is unusual and it is the reason the deployment is still current. A community bank does not have engineers sitting idle, so if changing the assistant’s wording requires a ticket and a release, the wording stops being changed and the assistant quietly drifts out of date.
Compliance, accuracy and customer data
A bank carries obligations that most organizations deploying an assistant do not, and three of them shaped what got built.
It answers, and it does not advise. A bank can tell you what a product does and what its terms say. It cannot tell you what to do with your money. Limiting retrieval to published material keeps the assistant on the right side of that line structurally, rather than by hoping a general model stays there.
Rates and terms have to be current or absent. A rate quoted from last quarter is worse than no rate at all, so the corpus tracks what the bank has actually published and the assistant declines when the published answer is not in it.
It has to be safe in front of anyone. A public assistant on a bank’s website takes every question, including ones about specific accounts it must never attempt to answer. The refusal behavior matters as much as the answers.
The capability that changed it: answers that do something
An assistant that only replies is a better FAQ. The change came when actions were switched on, so a question could end somewhere useful rather than in a paragraph: the right product page, the right application, the right branch.
Roughly one automated action now fires for every question asked. For a bank, that is the difference between deflecting a call and starting an account opening at eleven at night.
Two years in
| MEASURE | TO DATE |
| Consecutive days with a question answered | 330 |
| Questions this year against the year before | 5.6× |
| Questions answered to date | 14,586 |
| Answer rate, last six months | 91.6% |
| Support tickets raised in two years | 0 |
Set the last twelve months against the twelve before them and the assistant handled 5.6 times as many questions, climbing from roughly 180 a month to roughly 1,000. Nothing was relaunched and no campaign ran behind it. People used it, came back, and told others it was there.
The streak is what makes that growth credible. A question has been answered every day for 330 days and counting, weekends and holidays included, and nine in ten of them resolved from the bank’s own published material.
Then there is the number nobody thinks to ask for. Zero support tickets across two years of continuous operation. For a small team that owns the thing outright, an assistant that never needs rescuing is worth as much as one that answers well. The bank has renewed on enterprise terms and is now into its second enterprise term.
Why it held up
Grounding suits a regulated balance sheet. Own-content-only retrieval with honest refusal means a bank gets an assistant that declines cleanly, which matters more than one that always has something to say.
A marketing team can run it alone. Fifty-eight persona rewrites by the people who own the brand voice, with nothing between a decision and the live site.
And it keeps working without being looked after. Two years, 600 active days, zero support tickets, no migration and no rebuild. Which is why the conversation at renewal was a renewal rather than a re-evaluation.
What this means for your bank
Your customers are already asking. The site is answering with a menu.
The answer is usually already published. Rates, terms, eligibility and hours are on the site, approved and current. What is missing is a way in for somebody who does not know which page holds it.
Make sure the answer leads somewhere. An assistant that only replies saves a click. One that routes to the right application changes what the channel is worth to the bank.
And there is one question worth asking before you sign anything: who will own this in year two? If the answer is a project team, the deployment has a shelf life. This one is run by the function that owns the voice, which is why it is still current.
The customer at eleven at night now gets an answer instead of an opening time.
Arooj Ejaz is the Marketing Operations Lead at CustomGPT.ai, where she works on content, growth operations, and go-to-market programs for AI agent and chatbot solutions.