Table of Content
What are good examples of AI chatbots in business?
Good AI chatbot examples include Bank of America’s Erica for banking, Klarna’s assistant for customer service, Duolingo Max for language practice, Morgan Stanley’s internal assistant for advisors, and small-business tools such as Tidio’s Lyro and Intercom’s Fin. Each answers questions inside a narrow job, uses the company’s own information and hands off to a person when needed.
The short version
- The best chatbots do one job. Banking questions, order status, language practice and advisor research are all narrow, well-defined jobs.
- Scale is possible. Bank of America reports Erica passed 3 billion client interactions in August 2025, and Klarna reported 2.3 million conversations in its first month.
- Small businesses can start free. Tidio’s free plan covers 50 billable conversations, and HubSpot’s free CRM includes a chatbot builder.
- Wrong answers are your problem. In 2024 a tribunal held Air Canada responsible for bad advice its chatbot gave a customer.
- Rule-based and AI chatbots are different tools. Rules are predictable. AI handles open-ended questions, with more risk.
- Checked on October 13, 2026. Features and prices change, so confirm before you buy.
Types of chatbots in 60 seconds
A chatbot is software that talks with people in a chat window or by voice. Most chatbot examples fall into three types.
- Rule-based chatbots follow a script. A customer picks a button, and the bot follows a fixed path. They are predictable and cheap, and they fail on anything unexpected.
- AI chatbots use a language model to understand free-form questions and write answers from your content. They handle variety, and they need guardrails.
- Hybrid chatbots use rules for sensitive steps, such as refunds, and AI for open questions. Most business chatbots today work this way.
A chatbot that also takes actions in your systems, such as updating an order, is closer to an AI agent. We cover that line in our guide to AI customer service agents.
12 AI chatbot examples
Each example below is documented by the company, the vendor or a public record. Company numbers are self-reported, so we label them.
1. Bank of America Erica: banking help at huge scale
Erica is Bank of America’s virtual financial assistant, launched in 2018. In an August 20, 2025 press release, the bank said Erica had passed 3 billion client interactions, serves nearly 50 million users and handles more than 58 million interactions a month. It also reported that clients received more than 1.7 billion proactive, personalized insights from Erica, and that more than 98% of users find the information they need.
What it teaches: a narrow domain (banking) and years of improvement beat a general-purpose bot.
2. Klarna: a support assistant across many markets
Klarna’s AI assistant, built with OpenAI, handled 2.3 million conversations in its first month, according to a February 27, 2024 company release. Klarna said that was two-thirds of its customer service chats, that resolution time fell from 11 minutes to under 2 minutes, and that it worked in 23 markets and more than 35 languages.
What it teaches: multilingual coverage is one of the clearest wins for chatbots. The first-month figure is a company claim.
3. Air Canada: the cautionary chatbot example
In Moffatt v. Air Canada (2024 BCCRT 149), a British Columbia tribunal found Air Canada liable after its chatbot gave a customer wrong information about bereavement fares. The airline argued the chatbot was a separate legal entity. The tribunal rejected that, saying the chatbot is part of the airline’s website and that the airline did not take reasonable care to ensure its accuracy. Sources differ on the exact amount awarded.
What it teaches: your chatbot speaks for your business, and its answers can carry legal weight.
4. Morgan Stanley: an internal chatbot for advisors
The AI @ Morgan Stanley Assistant, built on GPT-4, lets wealth management advisors search the firm’s own research and documents in plain language. OpenAI’s case study reports that more than 98% of advisor teams actively use it.
What it teaches: internal chatbots carry lower risk than public ones and still save real time.
5. Duolingo Max: a chatbot as a language tutor
Duolingo launched Max in March 2023 with GPT-4 features. Explain My Answer lets learners chat with Duo, the owl, about why an answer was right or wrong. Roleplay lets learners practice conversations such as ordering coffee in a Paris cafe, with scenarios written by human experts and AI feedback afterward. Duolingo says it reviews generated content for accuracy and tone.
What it teaches: human-written scenarios plus AI conversation is safer than letting the model improvise everything.
6. Domino’s Dom: an early rule-based ordering bot
Domino’s launched Dom, a Facebook Messenger ordering bot, in August 2016. Early on, customers needed a saved Easy Order profile and the bot responded to little beyond the word “pizza”. In February 2017, Domino’s let customers order from the full menu through Messenger. We could not confirm whether Dom still runs today.
What it teaches: a rule-based bot can work well for one repeat task, such as reordering a favorite.
7. Tidio with Lyro: a chatbot built for small teams
Tidio pairs live chat with Lyro, an AI agent that answers customers from your content. The Free plan costs $0 with 50 billable conversations and 10 seats, and the first 50 Lyro conversations are free for life. Starter is $24.17 a month, and Growth starts at $49.17 a month.
What it teaches: a small store can launch a support chatbot in a day without a developer.
8. Intercom Fin: pay only for outcomes
Intercom’s Fin is priced at $0.99 per outcome. An outcome counts when a customer confirms the issue is resolved, stops asking for help after Fin responds, or Fin completes a workflow. Each conversation is charged once, and a minimum monthly commitment applies when Fin sits on top of an existing helpdesk.
What it teaches: pricing by outcome keeps cost tied to results, but read how the vendor defines an outcome.
9. HubSpot chatbot builder: a chatbot attached to your CRM
HubSpot’s free CRM lists a chatbot builder and live chat, alongside forms, email tracking and meeting scheduling, for up to two users and 1,000 contacts. A chatbot tied to the CRM can capture a lead and log it in one step.
What it teaches: a chatbot is most valuable when the answers and the lead data end up in the same system.
10. Zapier chatbots: an FAQ bot from your own documents
Zapier lets you add FAQs, documents and public links so a chatbot gives accurate, immediate answers, and connect it to other apps such as Zendesk for ticket responses.
What it teaches: a bot that only answers from approved documents is a safer first version than an open-ended one.
11. Shopify Sidekick: a chatbot inside the store admin
Shopify’s developer documentation describes Sidekick as the AI assistant built into Shopify that merchants use to get work done in the admin. Merchants ask in plain language, and Sidekick uses the store’s own data to answer and draft content.
What it teaches: chatbots for staff, not only customers, are a strong use case. See our guide to AI chatbots for e-commerce.
12. Microsoft Copilot Studio: build your own website chatbot
Copilot Studio lets teams build conversational and autonomous agents without code and publish them to websites, internal channels or Microsoft Copilot. Credit packs are $200 a month for 25,000 Copilot Credits, or you can pay as you go.
What it teaches: website chatbot examples can start as a configured product and grow into a custom build.
The 12 examples side by side
| Example | Type | Audience | Key fact |
|---|---|---|---|
| Bank of America Erica | AI assistant | Bank clients | 3 billion interactions |
| Klarna | AI support | Customers | 2.3M chats in month one |
| Air Canada | Website chatbot | Passengers | Held liable for wrong advice |
| Morgan Stanley | Internal AI | Advisors | 98% of teams use it |
| Duolingo Max | AI tutor | Learners | Roleplay, explain answers |
| Domino’s Dom | Rule-based | Customers | Launched August 2016 |
| Tidio Lyro | AI support | Small businesses | Free to 50 conversations |
| Intercom Fin | AI support | Helpdesk teams | $0.99 per outcome |
| HubSpot chatbot | Lead capture | Sales teams | Included in free CRM |
| Zapier chatbot | FAQ bot | Customers | Built from your documents |
| Shopify Sidekick | Staff assistant | Merchants | Works on store data |
| Copilot Studio | Builder | Any team | $200 for 25,000 credits |
Chatbot use cases by function
These chatbot use cases appear again and again. The example that proves each one is in brackets.
- Customer support. Answer order, billing and policy questions around the clock (Klarna, Tidio, Fin).
- Financial guidance. Balances, alerts and spending insights (Erica).
- Lead capture. Qualify visitors and log them in the CRM (HubSpot).
- Internal help desk. Answer staff questions from policies and research (Morgan Stanley, Zapier).
- Learning and coaching. Practice conversations and explain mistakes (Duolingo).
- Ordering and booking. Repeat orders and appointments (Domino’s).
- Store operations. Draft content and answer questions about store data (Sidekick).
How to pick the right chatbot
- List your top ten questions. If ten questions cover most of your volume, a simple FAQ bot may be enough.
- Decide what the bot may do. Answering is low risk. Refunds, payments and account changes need rules and approval.
- Limit it to approved content. Feed it your policies and pages, and tell it to say so when it does not know.
- Plan the handoff. Every chatbot needs a visible way to reach a person.
- Measure the right things. Track resolution, escalation, repeat contacts and customer ratings, not just conversation counts.
Mistakes that make chatbots fail
- No handoff. Customers stuck in a loop leave angry.
- Stale content. A bot trained on last year’s policy gives last year’s answers. Review the source documents on a schedule.
- Overpromising. Tell users they are talking to an AI and what it can and cannot do.
- Counting chats instead of outcomes. A high volume of unresolved chats is a cost, not a win.
- No testing on real questions. Test with a month of real tickets before launch.
Voice is a separate decision. Compare the two in our guide to AI chatbots vs AI voice agents.
Buy or build
Buy a chatbot when its features match your helpdesk and the content comes from public pages. Build when the chatbot must read order, account or ERP data, follow strict rules or run on your own channels. Our chatbot solutions team builds chatbots connected to your own systems, and a custom build can replace several overlapping subscriptions.
Frequently Asked Questions
What are examples of chatbots?
Examples include Bank of America’s Erica, Klarna’s customer service assistant, Duolingo Max, Morgan Stanley’s internal assistant, Tidio’s Lyro, Intercom’s Fin and Shopify’s Sidekick. They span banking, retail, education and e-commerce, and they range from scripted bots to AI assistants.
What is the best AI chatbot for a small business?
Tidio with Lyro is a strong starting point because its free plan covers 50 billable conversations. HubSpot’s free CRM includes a chatbot builder for lead capture. Choose by your main goal: support or lead capture.
What are common chatbot use cases?
Customer support, lead capture, internal help desks, ordering and booking, financial guidance and tutoring. The best use cases involve repeated questions, clear answers and an easy handoff to a person.
What is the difference between a rule-based and an AI chatbot?
A rule-based chatbot follows a fixed script and handles only paths you wrote. An AI chatbot uses a language model to understand open-ended questions and write answers from your content. AI is more flexible and needs more guardrails.
Can a chatbot be wrong, and who is responsible?
Yes, and you are responsible. In Moffatt v. Air Canada, a tribunal held the airline liable for wrong information its chatbot gave about bereavement fares. Limit the bot to approved content and review its answers regularly.
How much does a business chatbot cost?
Small-business chatbots start free: Tidio’s free plan covers 50 conversations and Starter is $24.17 a month. Intercom’s Fin costs $0.99 per outcome. Custom chatbots are quoted by scope and integrations.
What are good website chatbot examples?
Tidio’s Lyro and Intercom’s Fin run on business websites, and Copilot Studio lets teams build and publish their own. Air Canada’s chatbot is a reminder to test any website chatbot carefully before launch.
Will a chatbot replace my support team?
It takes over repeated questions, not hard cases. Klarna reported its assistant handled two-thirds of chats in its first month, and complex issues still need people. Plan for a handoff and keep humans for exceptions.
How long does it take to launch a chatbot?
A simple FAQ chatbot can launch in a day on a no-code tool. A custom chatbot connected to your order, CRM or ERP data takes longer. Test on real questions before you go live.
Bottom Line
The strongest chatbot examples share a pattern: a narrow job, approved content, honest limits and an easy path to a person. Erica and Klarna show what scale looks like, Air Canada shows the risk, and Tidio and HubSpot show how a small business can start free.
Pick one job, feed the bot your best content and measure outcomes. When you want a chatbot that reads your own data, our chatbot solutions team can scope it with you.
