How to Make Money Building AI Chatbots for Businesses

You can make money with AI bots by building and maintaining chatbots for small businesses: a website or messaging bot that answers customer questions from the business's own information, captures leads or books appointments. Most people who earn from AI chatbots charge a one-time setup fee and a monthly fee to keep the bot accurate, using no-code platforms rather than writing code.

It's a real service, not a shortcut. The platforms are easy to start with, which means competition is easy too. What a business pays for is a bot that gives correct answers, knows when to hand over to a person and stays up to date when prices or policies change.

How do you make money with AI chatbots?

You sell a narrow, reliable bot to a specific type of business, then get paid to look after it. The bot is built on a platform the client pays for; you provide the setup, content, testing and upkeep.

A typical engagement has three parts:

1. Discovery. Learn the top questions customers ask, what the business wants the bot to do and what it must never say.
2. Build and test. Load approved content, write the bot's instructions, connect it to the website or booking tool and test it against real questions.
3. Maintenance. Review conversations, fix wrong answers, update content and report on what customers ask.

That maps onto the broader method in our guide to making money with AI by selling a service: pick one client type, build proof, package one offer. Chatbots sit next to workflow automation; if you want to offer both, see AI automations small businesses will pay for.

What kind of chatbot will a small business pay for?

Bots that answer repetitive questions or capture revenue the business is currently missing. "An AI assistant for your website" is vague; "a bot that answers after-hours questions and books consultations" is a sale.

Start with one type and one industry. A bot built for dental clinics can be adapted for the next clinic much faster than starting fresh, and your demo becomes more convincing.

Which no-code chatbot platforms should you use?

Choose a platform that can stick to the business's own content, hand off to a human and export conversation logs. Price matters, but reliability and control matter more.

Prices below are the public list prices as of 2026; check each page before quoting a client, because plans and limits change often.

Who should pay for the platform?

The client, on an account in their name. That way they own their data and conversation history, you aren't stuck paying for a bot after a client leaves, and a handover is simple. Ask for collaborator access instead of sharing passwords.

What about building with an AI API directly?

Developers can build custom bots on a model provider's API, which gives more control but adds hosting, security and maintenance work. For a first client, a no-code platform is usually the faster and safer choice.

How do you stop a chatbot from giving wrong answers?

Limit what the bot knows, tell it what not to do and test it like a skeptical customer. You can reduce mistakes, not remove them, so plan for the ones that get through.

The legal backdrop is simple: the business owns what its bot says. In Moffatt v. Air Canada, decided in 2024, a British Columbia tribunal held the airline liable after its chatbot gave a customer incorrect information about bereavement fares, and rejected the idea that the chatbot was responsible for its own statements.

Build every bot with these controls:

• Approved sources only. Train it on the business's current website pages, policies and FAQ, not on general knowledge.
• Clear boundaries. Instruct it not to quote prices, discounts, medical, legal or financial advice unless that exact information is in its sources.
• Human handoff. When it isn't sure, it should say so and offer a phone number, email or live chat.
• A test set. Write 30 to 50 real customer questions, including awkward and off-topic ones, and check every answer before launch.
• Visible limits. A short line in the chat window explaining it's an AI assistant and how to reach a person.

Example (hypothetical): a physiotherapy clinic's bot is asked whether a treatment is covered by a particular insurer. A safe bot answers that coverage depends on the plan and offers to pass the question to the front desk. An unsafe bot guesses.

What privacy rules apply to customer data in a chatbot?

Collect as little as possible, tell people what happens to it and only use tools whose data terms the client has approved. Chatbot conversations often contain names, phone numbers and personal details people type without being asked.

Data the bot handles

• Ask only for what's needed to reply or book, typically a name and one contact method.
• Tell the bot not to request sensitive information such as health details, payment card numbers or ID numbers.
• Update the privacy notice on the client's website to mention the chat tool and what it stores.
• Check the platform's data terms. Look for where data is stored, how long logs are kept and whether a data processing agreement is available. OpenAI, for example, says it doesn't train on API or business-plan data by default; platforms built on top of model providers set their own terms.

Which privacy laws apply depends on where the business and its customers are. In the US that may include state privacy laws; the UK and EU have GDPR; Canada has PIPEDA and provincial laws. Regulated businesses such as clinics may have stricter rules, so the client should check with their own adviser.

Telling people they're talking to a bot

• California. Business and Professions Code section 17941 makes it unlawful to use a bot online to mislead people about its artificial identity in order to drive a sale, and provides that clear, conspicuous disclosure avoids liability.
• EU. The EU AI Act's transparency obligations, which apply from August 2, 2026, require chatbots to be designed so users are told they're interacting with an AI system unless that's obvious.

The simple rule: always label the bot as AI, everywhere.

How should you price and write a chatbot maintenance contract?

Charge a setup fee for discovery, build and testing, and a monthly fee for a defined maintenance scope. Write down what the monthly fee covers so it doesn't turn into unlimited support.

A maintenance contract should state:

1. Review cadence. For example, reading flagged and unanswered conversations every week.
2. Content updates. How many changes to prices, policies or services are included each month.
3. Response time. How quickly you fix a wrong or harmful answer once it's reported.
4. Reporting. A short monthly summary of top questions, handoffs and gaps in the content.
5. Platform costs. That the client pays the platform directly, and what happens if usage exceeds the plan.
6. Responsibilities. That the client approves the bot's content and remains responsible for its business information, and that you aren't liable for answers outside the approved scope.
7. Ending the contract. Notice period, handover of documentation and removal of your access.

Example (hypothetical): if discovery, building and testing a FAQ bot takes you about 10 hours and your floor rate is $40 an hour, your setup minimum is $400, before any value-based adjustment. The monthly fee would cover your estimated review and update hours. Your numbers will depend on your market and the bot's complexity; our guide on how to price AI services shows how to calculate the floor and anchor to value.

Never promise a resolution rate, more sales or fewer support calls. Promise the build, the testing and the maintenance you'll do.

How do you find your first chatbot clients?

Show a business what its own bot could look like, using only its public information, and keep the demo private. A working demo explains the offer faster than any pitch.

• Pick one niche where customers ask the same questions often, such as clinics, gyms or home services.
• Build a private demo from a prospect's public website, then share a short screen recording. Don't publish or deploy a bot for a business without its permission.
• Point out the gap you noticed, such as questions left unanswered after hours.
• Offer a small first project: one bot, one channel, a fixed setup scope.

Outreach and packaging work the same way as for other AI offers; our list of AI services you can sell today covers first-client outreach and anti-spam rules.

Next steps

Choose one niche and build a test bot on a free plan using a mock business's FAQ, then write your 30-question test set and your maintenance checklist. When you're ready to package it and find clients, our step-by-step guide to making money with AI walks through the full path. You can also explore our guide to making money online to compare other options.