AI chatbots for small business, across web and messaging

It is a conversational system that uses approved business information to answer routine questions, collect enquiry details and trigger agreed next steps such as a booking, notification or human handoff. It can live on a website and, where platform access allows, selected messaging channels.

Potentially. Website chat, WhatsApp, Facebook Messenger, Instagram direct messages and SMS use different business accounts, APIs and policies. We confirm which channels are technically available, appropriate and included before we scope the build.

Yes, when a suitable calendar or booking platform is connected and the booking rules are clear. The chatbot can show approved availability, collect required details and confirm the appointment. Complex suitability or priority decisions should stay with a person.

Any generative AI system can produce an incorrect answer, so no responsible provider should promise zero mistakes. We reduce the risk with an approved knowledge source, narrow scope, explicit do-not-answer rules, testing, monitoring and a handoff whenever the system is uncertain.

It should say that it is unsure, collect only the details needed for follow-up and route the conversation to the nominated person or queue. The handoff rule is designed before launch rather than improvised after a failure.

It is best used for repetitive questions, initial capture, routing and simple connected actions. Staff still own judgement, emotional conversations, complaints, sensitive advice, exceptions and the ongoing quality of the customer experience.

Usually, if the CRM provides a suitable integration or API. HubSpot, Zoho and other established systems can receive contact details, tags, summaries or tasks depending on their plan and the agreed build. We verify the real connection before including it in scope.

Useful reporting covers conversation volume, common question themes, contact details captured, bookings or other completed actions, human handoffs, unanswered questions and failed connections. Those signals are more useful than a raw count of messages.

Timing depends on the number of channels, quality of the business information, booking and CRM integrations, approval requirements and testing depth. A written scope should separate content preparation, build, platform approval, testing and launch rather than promise an instant result.

Cost depends on channels, conversation volume, integrations, knowledge preparation, testing, monitoring and ongoing support. LUNA provides a custom written scope after discovery instead of publishing a package that assumes every business needs the same chatbot.

Only the minimum information needed for the next legitimate step. A routine enquiry may need a name, contact method, service and preferred time. Sensitive details should not be collected simply because the chatbot can ask for them.

It can help with approved factual information, administrative questions and booking, but it should not diagnose, provide clinical advice or decide whether treatment is suitable. Healthcare practices need stricter content, privacy, escalation and human-review rules.

Yes. Chat should add a useful contact route, not trap every customer inside automation. Keep phone and email options visible and provide a clear human handoff. Businesses that also need phone coverage can use the separate AI receptionist service.