A banking chatbot is software that answers customer or member questions in a bank's website, app, or online banking, and in more advanced versions completes routine tasks like checking a balance or locking a card. Most large U.S. banks already run one. The hard part is not the technology. It is deploying it in a way that meets consumer protection rules, protects account data, and gets people to a human quickly when the bot is out of its depth.
This guide covers the use cases that work, the regulatory issues regulators have already flagged, a checklist for evaluating vendors, and a short list of vendors that build specifically for financial institutions. It is general information, not legal or compliance advice. Your compliance team and counsel should review any deployment.
How common are banking chatbots?
Very. In its June 2023 issue spotlight "Chatbots in consumer finance," the Consumer Financial Protection Bureau (CFPB) found that each of the 10 largest U.S. commercial banks had deployed chatbots in customer service, and estimated that roughly 37% of the U.S. population interacted with a bank's chatbot in 2022.
The best-known example is Bank of America's Erica. In a March 2026 newsroom release, Bank of America said 20.6 million users interacted with Erica nearly 700 million times in 2025, and that total client interactions since its 2018 launch have passed 3.2 billion. Most institutions will never operate at that scale, but it shows how much routine service volume a well-built assistant can absorb.
Banking chatbot use cases
The use cases below are ordered roughly from lowest to highest risk. Most institutions start at the top and work down as their controls mature.
Public information and FAQs
Branch and ATM locations, hours, routing numbers, how to order checks, what documents you need to open an account, and how to reach a specific department. This is pre-login, low-risk, and high-volume, and it is where most credit unions and community banks begin.
Product questions and lead capture
Explaining the differences between checking accounts, how a HELOC works, or what a certificate term means, using only your published product information. The bot can then capture the person's interest and contact details for a loan officer or member services rep, or book an appointment. For fintech companies, this is often the main job of the website chatbot.
Authenticated account servicing
Once a customer logs in, a chatbot integrated with online banking can show balances, recent transactions, and statements, or help with card activation and locking. This requires strong authentication, tight integration with your core or digital banking provider, and careful logging.
Payments and transfers
Some assistants can initiate transfers or bill payments inside the authenticated session. This raises the stakes considerably: every action needs confirmation, limits, and a clear audit trail.
Disputes, fraud, and hardship
Reports of unauthorized transactions, billing errors, and financial hardship carry specific legal obligations and real harm if mishandled. Most institutions treat these as topics the bot should recognize quickly and route to a trained person, not resolve on its own. The CFPB's report specifically warns about chatbots that only recognize a dispute when the customer uses certain words, that repeat back the same account information the customer is disputing, or that make it hard to reach timely human help.
Employee assistance
Some platforms also offer assistants for contact center staff, surfacing policy answers or summarizing member history so a human agent can resolve issues faster. This keeps a person in the loop while still reducing handle time.
Regulatory considerations for banking chatbots
No single rule governs banking chatbots. Instead, the rules that apply to your customer service apply to the chatbot too. The main themes regulators have raised:
Consumer financial protection
The CFPB's 2023 issue spotlight states plainly that chatbots must comply with all applicable federal consumer financial laws, and that institutions may be liable when they do not. It flags risks including inaccurate information, failure to solve complex problems, customers getting stuck in loops without reaching a human, and harm to customer trust. Design your bot so it gives accurate answers from approved content, recognizes when a request is outside its scope, and offers a clear path to a person.
Third-party risk management
If a vendor provides your chatbot, it is a third-party relationship. The Federal Reserve, FDIC, and OCC issued final Interagency Guidance on Third-Party Relationships: Risk Management in June 2023. It says using a third party does not diminish a banking organization's responsibility to operate safely and comply with applicable laws, including consumer protection laws and prohibitions on unfair, deceptive, or abusive acts or practices. Expect to run due diligence, negotiate contract terms, and monitor the vendor over time, scaled to how critical the activity is.
Privacy and data security
Customer financial information is protected under federal privacy law and your own security program. Decide what data the chatbot may see, where transcripts are stored, how long they are kept, and whether any data leaves your environment for an AI model provider. The CFPB's report also highlights security risks, including fake impersonation chatbots used for phishing and personal data entered into chat logs becoming another target for attackers.
Authentication and fraud
Anything account-specific should happen only after proper authentication, ideally inside your online banking session rather than on the public website. Watch for social engineering, where someone tries to get the bot to reveal information or take actions without being verified.
Fair treatment and accessibility
Make sure the chatbot works for customers with disabilities, for people with limited English proficiency where you serve them, and does not steer people toward or away from products in ways that create fair lending concerns.
A checklist for evaluating banking chatbot vendors
Take these questions into every vendor conversation and get the answers in writing.
- Scope: Which tasks does the bot handle automatically, and which does it always route to a human? Can you configure that per topic?
- Grounding: Does it answer only from your approved content, and can it cite the source?
- Integrations: Which core, digital banking, and CRM systems does it connect to, and does account access happen inside an authenticated session?
- Security evidence: Which independent audits or reports can the vendor share, such as a SOC 2 Type II report? Ask for the report itself.
- AI model use: Which model providers process your data, where, and under what terms? Is your data used for training?
- Data retention: How long are transcripts kept, and can you set and enforce deletion?
- Audit trail: Can you see every conversation, every action taken, and who approved it?
- Handoff: How quickly and how cleanly does a conversation reach a human, and does the human see the full context?
- Complaint handling: Does the bot recognize complaints and disputes and route them appropriately?
- Exit: Can you export your data and content if you leave?
How we chose the vendors below
We looked for vendors whose own websites show a product built specifically for banks, credit unions, or financial services, with conversational AI as a core offering. We describe each product from its own site and attribute any claims to the vendor. We did not test these products, and none of this is an endorsement. We did not find published list prices for these vendors, so pricing is listed as "see vendor site" as of September 2026.
Notable banking chatbot vendors
Kasisto (KAI)
- Best for: banks that want a banking-specific conversational AI platform integrated with their digital banking provider.
- Key features: the KAI platform with a banking-specific language model (KAI-GPT), a no-code content portal, integration with live chat and contact center systems, and access through digital banking partners including NCR, FIS, and Q2.
- Pricing: see vendor site.
- Limitations: Kasisto's site announces that it has been acquired by Backbase, so check how the roadmap and contracts will be handled going forward.
Glia
- Best for: regional and community banks and credit unions that want AI, digital, and voice service on one platform.
- Key features: AI self-service across banking journeys, voice AI, chat, video, SMS, and secure messaging with context carried between channels, agent assist, and integrations with online banking platforms and cores. Glia says it works only with financial institutions.
- Pricing: see vendor site.
- Limitations: a broad contact center platform. If you only want a website FAQ bot, it may be more than you need.
interface.ai
- Best for: credit unions and community banks that want voice and chat AI connected to core systems.
- Key features: agentic voice AI for phone banking, a chat assistant for website and mobile app, risk-based multi-factor authentication including voice biometrics, and prebuilt integrations with core banking systems, CRMs, and payment gateways.
- Pricing: see vendor site.
- Limitations: voice-first positioning. Evaluate the chat product on its own if chat is your main need.
Posh
- Best for: credit unions and community banks that want a website digital assistant plus AI search and voice.
- Key features: a Digital Assistant chatbot for websites and mobile apps that answers questions and handles routine tasks, Posh Answers (AI-powered site search), a voice assistant, handoff to live chat with full context, and a central portal for managing content across channels.
- Pricing: see vendor site.
- Limitations: transactional features depend on integrations with your banking systems, so confirm which ones are supported for your core.
Where a general-purpose chatbot fits
Many fintech companies, smaller credit unions, and lenders mainly need help with the public, pre-login side of the website: answering product questions from published information, capturing leads, and booking appointments with a loan officer or advisor. A general-purpose AI agent can do that job, as long as it never touches account data.
Bund AI is a general-purpose web chat and email agent, and we want to be precise about where it fits in banking. It can:
- Answer general questions from your published content, such as account types, how to apply, required documents, branch hours, and how to contact a department.
- Capture a prospective customer's name, contact details, and interest as a lead, with the qualifying questions you choose. See lead capture.
- Book an appointment with a loan officer or advisor against real calendar availability. See appointment booking.
- Hand the conversation to a person with the full transcript when a question needs one.
It should not be used for regulated account servicing. Bund AI does not integrate with core banking systems out of the box, is not designed for authenticated account access, transactions, disputes, or fraud reports, and does not hold SOC 2, PCI DSS, or other security certifications. Card numbers entered in a conversation are masked at storage, but that is not a substitute for a platform built and audited for financial services. For authenticated banking tasks, use a vendor that specializes in financial institutions and can provide the security evidence your third-party risk program requires.
If you want to understand the difference between a scripted FAQ bot and an agent that can take actions, read our explainer on chatbots versus AI agents. For guidance on getting escalation right, see when AI should hand off to a human. Plan details are on the pricing page.
Frequently asked questions
What can a banking chatbot do?
Common uses include answering FAQs about branches, hours, and products, capturing leads and booking appointments, and, inside an authenticated session, showing balances and transactions, activating or locking cards, and sometimes initiating transfers. Disputes, fraud, and hardship requests are usually routed to trained staff.
Are banking chatbots regulated?
There is no chatbot-specific federal rule, but the laws that govern your customer service apply to the chatbot. The CFPB has said chatbots must comply with federal consumer financial laws, and interagency guidance from the Federal Reserve, FDIC, and OCC says using a third-party vendor does not reduce a bank's responsibility for compliance.
What are the biggest risks of a banking chatbot?
The CFPB has highlighted inaccurate answers, failing to recognize complex problems, making it hard to reach a human, and privacy and security risks. Poor handling of disputes or complaints is a particular concern because those carry specific legal obligations.
Can a credit union use a general AI chatbot on its website?
For public, pre-login questions and lead capture, often yes, subject to your own vendor risk review. For anything involving member account data, authentication, or transactions, use a vendor built for financial institutions that can provide the security documentation your examiners will expect.
Is Bund AI suitable for banks?
Only for non-regulated, pre-login use, such as answering general product questions from your published content, capturing leads, and booking appointments. Bund AI does not integrate with core banking systems, is not built for authenticated account access or transactions, and does not hold SOC 2 or PCI certifications.