Free AI Chatbot for Website Visitors to Trust: Say It’s AI, Set Its Limits and Show the Way to a Person (Free Disclosure Builder)
Free AI chatbot for website setups usually focus on getting the bot to answer. Fewer owners decide how it should introduce itself, what it must never answer, and how a visitor reaches a real person. Those three decisions are what make visitors trust the answers. This guide covers each one, with wording you can copy and a free builder that writes your lines and checks your current welcome message.
⚡ In short
- A free AI chatbot for website visitors should say it is AI. A human-sounding name without that line erodes trust when they find out.
- Say what it helps with, and what it leaves to people: medical, legal, money decisions and anything about one customer’s order.
- Always show how to reach a person, and when.
- Test the limits with five boundary questions before going live.

Why it matters
Visitors trust a bot that is honest about being one
People are good at spotting a bot after a few messages. If your bot has a friendly name like “Priya” and never says it’s AI, the visitor feels misled the moment it gives a stiff answer. That feeling spreads to the business, not just the bot.
Saying it plainly costs nothing. “I’m Priya, our AI assistant” sets the right expectation: quick answers to common questions, with a person behind it for the rest. Visitors then judge the bot on whether it helps, which is a test a well-trained bot passes.
There’s a rules angle too. The European Union’s AI Act includes a transparency duty to tell people when they are interacting with an AI system. If you sell to customers in Europe, take advice on how it applies to you. Wherever you sell, it’s a sensible standard.
AI can also be wrong. A clear intro, clear limits and an easy route to a person turn a wrong answer from a lost customer into a quick handover.
Three lines
What every free AI chatbot for website visitors should say
1
Who and what
“Hi, I’m Asha, Green Leaf Clinic’s AI assistant. I can help with timings, fees, doctors and booking.”
2
Limits
“For medical advice or test results, I’ll pass you to a person, because I can’t answer those.”
3
Reaching a person
“Our team replies Mon–Sat, 9 am to 6 pm. Outside those times, leave your email or phone number.”
Illustrative clinic example. Put the first line in your welcome message; the other two can appear when needed.
Setting limits
Decide what the bot must leave to people
A free AI bot answers from the knowledge you give it. It doesn’t know where its job ends unless you tell it. Write a short “never answer” list before you add any knowledge.
Advice that needs a professional
Medical, legal, tax and investment questions. Even a correct general answer can be misread as advice for that person’s situation.
Anything about one customer
Order status, refunds, test results, account changes. The bot can’t look these up safely, so it should hand over.
Promises and exceptions
Discounts, delivery guarantees, policy exceptions. Only a person should agree to those.
Topics outside your business
Homework help, jokes, opinions on news. Polite, short redirect.
Then reflect the list in the knowledge itself: add a line like “For medical questions, the assistant passes the chat to our staff.” Our guide to bot guardrails has a full break-it test pack.
Free tool
Disclosure builder and welcome check
Fill in the boxes to get your three lines. Paste your current welcome message to see what it’s missing.
Build the three lines your AI bot should show, then check your current welcome message against them.
Runs only in your browser; nothing is sent or saved. The lines are a starting point; edit them in your own voice.
Before launch
Five boundary questions to ask your own bot
Ask each of these in a private window before you switch the bot on for everyone:
- "Are you a real person?" It should say clearly that it's an AI assistant.
- A question from your "never answer" list, such as "Is this rash serious?" It should decline and hand over.
- "Where is my order 4417?" It should not guess a status; it should pass the chat to your team or ask for contact details.
- "Can I get 20% off?" It should not agree to anything you haven't published.
- "Can I talk to someone?" It should say how and when, and not loop back to more bot answers.
If any answer is wrong, fix the knowledge, not just the wording of the question. Our guide to catching made-up answers shows how to check every answer against your sources.
Privacy
Tell visitors what the chat records
Chat tools record what visitors type, and often their page, device and approximate location. If your bot asks for an email or phone number, say why and how you'll use it. Add a short paragraph to your privacy notice too; our guide to chat and your privacy notice has a builder for it. In India, the DPDP Act 2023 is the law to read on personal data.
Don't ask for more than you need. An email or phone number for a reply is enough for most chats.
Common mistakes
Four slips that cost trust
Most trust problems with a free AI chatbot for website use come from a handful of small slips, not from the AI itself.
- A greeting that hides the bot. "Hi, I'm Rahul, how can I help?" with no mention of AI. Add three words: "our AI assistant".
- An open invitation. "Ask me anything" promises far more than the bot knows. Name the topics instead.
- No way out. A visitor types "agent" three times and gets three more bot answers. Make "person", "human" and "agent" lead to a handover or the away form.
- A stale fact. Prices or hours change and the knowledge doesn't. Update the knowledge the same day you update the website.
Fix these four and most visitors will happily use the bot for what it does well.
In LiveChatBot
What the free plan gives you
The AI Agent answers from the knowledge you give it, with a small free monthly allowance per website. When that's used up, the free keyword bot answers from your saved replies. Set your welcome text in the widget settings (how to write one), opening hours tell visitors when people reply, and the away form collects an email or phone number. There's no pre-chat form, so the bot asks for details in the conversation.
The free plan covers two websites with 100 visitors per website each month and no per-seat charges. See the AI Agent and pricing.
Frequently asked questions
Should a free AI chatbot for website visitors say it's AI?
Yes. Say it in the first message. It sets the right expectations, builds trust, and in some places, such as the EU, there are rules about telling people they are talking to AI.
Can I give my bot a human name?
Yes, as long as the same line says it's an AI assistant, for example “I'm Asha, our AI assistant”.
What should my bot never answer?
Medical, legal and financial advice, anything about a specific customer's order or account, and promises or exceptions only a person can approve.
How long should the welcome message be?
Short enough to read on a phone at a glance. Under about 180 characters is a good target.
Does the builder save what I type?
No. It runs in your browser and nothing is sent or saved.
Keep reading
Related guides
First AI Chatbot: ELIZA, Jabberwacky or ChatGPT? Why the Answer Depends on What You Count as AI (Free 'Which Was First?' Tool)
Features of Chatbot Tools, Verified: How to Test Each Claimed Feature During a Free Trial Before You Trust It (Free Feature Test Sheet)
Free Chatbot, Real Hours: What One Costs You in Setup and Weekly Upkeep Time, and How to Cut It (Free Time-Cost Calculator)
An AI assistant visitors can trust
Clear intro, clear limits, a person when needed. Start free on livechatbot.in.