AI Chatbot or Rule-Based Chatbot: Which Does Your Business Need?
Rule-based bots follow scripts. AI assistants understand free-form questions. Each suits different jobs, and many businesses benefit from combining them.
3 min read · Evergreen guideLaunching a Customer-Facing AI Assistant: Scope, Guardrails and Handover
A good launch depends less on the AI and more on the decisions around it: what it may answer, what it must never do and when a person steps in.
Key takeaways
- Define in writing what the assistant may and may not answer.
- Test with real customer questions, including awkward and hostile ones.
- Make handover to a person easy, and tell the customer when it happens.
- Review conversations weekly at first, and fix the content, not just the bot.
1. Decide the scope
Write down the jobs the assistant will do: answer product questions, share opening hours, capture enquiries, book appointments. Then write what it must not do: give medical, legal or financial advice, promise prices or delivery dates it cannot check, discuss competitors or handle payments in chat.
2. Prepare approved content
The assistant is only as accurate as the information it can draw on.
- Collect current pages, brochures, price information, policies and FAQs.
- Remove outdated and contradictory versions.
- Assign one person to own each topic and update it when things change.
- Write short, clear answers for the top questions in your own voice.
3. Set guardrails
Guardrails are rules that keep the assistant safe and on topic.
- Answer only from approved content, and admit when it does not know.
- Decline off-topic or risky requests politely.
- Never ask for sensitive details such as passwords or full card numbers.
- Make clear that the customer is talking to an automated assistant.
4. Design the handover
- Offer a visible way to reach a person at any time.
- Trigger handover automatically on frustration, repeated misunderstanding or sensitive topics.
- Pass along the conversation so the customer does not repeat themselves.
- Say what will happen next and when.
5. Test properly
Build a test set from real questions from emails, calls and messages. Include misspellings, other languages, vague questions, trick questions and rude messages. Score each answer as correct, partly correct or wrong, and fix the cause: missing content, unclear content or a rule gap. Re-run the whole set after every change.
6. Protect privacy
Tell customers what is stored and why, collect only what you need, restrict who can read conversations and delete data you no longer need. Check the rules that apply to messaging on your channels.
7. Launch in stages
Start with a small audience or a single page. Review conversations daily in the first week, weekly afterwards. Track the share of questions answered, handovers, lead quality and customer feedback.
8. Keep it current
Update content when prices, policies or products change. Add answers for new questions that appear in the logs. Re-test regularly.
Our AI Chatbot & WhatsApp Assistant service covers each of these steps, and our Knowledge Assistant suits larger document collections.
Frequently asked questions
How long does it take to launch?
A focused assistant can launch within a few weeks. Preparing accurate content and testing usually take longer than the technical build.
What if the assistant gives a wrong answer?
Fix the underlying content or rule, add the question to your test set and re-test. Logging every conversation makes these problems visible quickly.
