AI chatbots and agents for small businesses: what they can do, what they cannot and where to start

Chatbot, agent, receptionist: what is the difference?

The words get used loosely, so it helps to separate them. A chatbot answers questions in a chat window, usually from a set of information you give it. An AI agent goes a step further: it can take actions, such as looking up a record, booking a slot or sending a follow-up, within limits you define. A voice agent or AI receptionist does the same job over the phone.

None of these is a person, and none should pretend to be. The useful question is not 'which is smartest' but 'which job do I want taken off my team's plate'.

Jobs they handle well

AI works best on work that is frequent, predictable and low-risk:

• Answering the same questions about services, opening hours, location, process and typical timelines.
• Capturing an enquiry properly: name, contact details, what the person needs and how urgent it is.
• Booking, rescheduling and reminding, when it is connected to your calendar.
• Qualifying leads, so your team speaks first to the enquiries that fit.
• Searching your own documents, policies or past work to give staff a quick answer.
• Covering evenings and weekends, when nobody is at the desk.

Jobs they should not do alone

Some work belongs with a person, however good the technology is:

• Legal, medical, financial or other regulated advice.
• Pricing exceptions, refunds, complaints and anything emotionally charged.
• Decisions that are hard to reverse.
• Anything where a wrong answer is costly and the assistant cannot check it.

Why it must be grounded in your information

A general-purpose AI will happily produce a confident answer that is wrong. A business assistant should answer from your approved content: your service pages, price ranges, policies and FAQs. When the answer is not in that material, the right behaviour is to say so and hand over to a person, not to guess.

This is why the setup work matters more than the model. A short, accurate, well-organised knowledge base produces a better assistant than a clever one fed with vague information.

Privacy and honesty

Tell people they are talking to an automated assistant, and always offer a route to a human. Collect only what you need for the task, and be clear about where data is stored and who can see it. If you serve clients in the EU, UK or US healthcare and similar fields, check the rules that apply to you, such as GDPR or HIPAA, before you switch anything on.

A safe way to start

Begin small, learn quickly and widen the scope as trust grows:

• Pick one job, such as after-hours enquiries or booking.
• Write down the 15 to 20 questions people ask most, with the answers you want given.
• Run it alongside your team for a few weeks and read every conversation.
• Fix the gaps, then add the next job.
• Review monthly: what it handled, what it handed over and what it got wrong.

What it costs and how long it takes

A focused chatbot or receptionist is usually a matter of weeks, not months, and the main cost is setup and ongoing tuning rather than the software itself. Costs depend on the channels (website, phone, WhatsApp), the systems it connects to and the volume. We give a fixed scope and price after a short call, so you know what you are buying before any work starts.

How to tell if it is working

Judge it on business results, not on how impressive it sounds:

• Enquiries captured out of hours that would previously have been lost.
• Time your team no longer spends on repeat questions.
• Share of conversations resolved without a handover, and how many handovers were appropriate.
• Feedback from the people who used it.