Build an AI Chatbot for a Local Business in One Afternoon

Most local business websites answer the same ten questions every single day — and lose the customers who ask them at 11 PM. An AI chatbot for local business owners changes that math: it answers instantly, books appointments, and collects leads around the clock, without hiring extra staff. In this tutorial, you’ll build one in a single afternoon with a no-code chatbot builder — and then package it as a paid service with a setup fee plus a monthly retainer.

Why an AI chatbot for local business is a great first AI service

Chatbots are one of the easiest AI services to sell because the value is obvious in a 60-second demo. A plumber’s site that answers “Do you do emergency calls?” at midnight is worth more than any slide deck. As a freelancer, you get three advantages: delivery costs are near zero after the first build, every local niche (dentists, roofers, salons, HVAC) needs the same core features, and the monthly retainer gives you recurring revenue. One afternoon of work can turn into a client who pays you every month.

Step 0: Pick your no-code chatbot builder

You have three solid options. Pick one and learn it deeply instead of dabbling in all three.

  • Chatbase — Train a bot on a website URL, PDFs, and text snippets, then embed it as a chat widget on any site. Lead-capture forms and human handoff are built in. A free tier exists for building and testing your first bot.
  • Tidio (with Lyro AI) — Best when the client also wants live chat. The AI agent answers common questions and hands off to a human when it can’t. Strong fit for shops and clinics that already have staff answering messages.
  • ChatGPT custom GPTs (fallback/demo option) — Free to prototype and easy to share a link for demos, but harder to embed cleanly on a client’s website. Use it to mock up a bot fast, then rebuild in Chatbase or Tidio for the production version.

For this tutorial, the steps assume Chatbase, but the same workflow applies to any builder.

The one-afternoon build: 5 steps

Step 1: Gather the raw material (about 60 minutes)

A chatbot is only as good as what it knows. Collect these from the client or their website:

  1. Full services list with prices or price ranges
  2. Business hours, address, parking info, service area
  3. Booking or contact links (Calendly, online scheduler, phone)
  4. Insurance accepted, payment methods, deposit/cancellation policy
  5. The 10 questions they answer repeatedly (ask them — they know)

Concrete example — a dental clinic: services (cleanings, whitening, Invisalign-style aligners), accepted insurance plans, new-patient special, emergency protocol (“Call this number; don’t use chat for emergencies”), parking garage info, and a link to book a checkup. Write each answer in plain language, one fact per line. If a fact isn’t documented, the bot will guess — so document everything.

Step 2: Train the bot (about 30 minutes)

In Chatbase, create a new chatbot and feed it the material:

  1. Add the client’s website URL as a data source so the bot crawls public pages (services, FAQ, contact).
  2. Upload PDFs or documents — price lists, patient intake forms, policy sheets.
  3. Paste short Q&A pairs as text snippets for the questions you collected in Step 1. Snippets are your fastest lever: whenever the bot answers poorly later, you fix it by adding a snippet, not by retraining everything.

Step 3: Write the behavior brief (about 30 minutes)

Under Build → Instructions, tell the bot how to act. Here’s a template you can adapt:

“You are the front-desk assistant for [Business Name], a [plumber/dentist/etc.] in [City]. Answer in a friendly, plain-spoken tone. Use only the information in your knowledge base — never invent prices, hours, or services. If a question is outside your knowledge, say so and offer the phone number or booking link. For emergencies (flooding, gas smell), tell the caller to call [number] immediately. Always end booking-related answers with the booking link.”

Concrete example — a plumber: add rules like “For ‘how much does X cost?’, give the typical price range from the knowledge base and say final quotes require a visit” and “Offer the emergency line for burst pipes, and the booking form for everything else.” This brief is what separates a professional bot from a generic one — and it’s a line item you charge for.

Step 4: Test it like a customer (about 45 minutes)

Open the playground and ask 15–20 questions as if you were a real customer, including edge cases: pricing questions, “do you offer discounts?”, rude messages, and questions the business can’t answer (“Do you do commercial work?” when they don’t). Watch for three failure modes: invented facts, tone that doesn’t match the brand, and dead ends with no next step. Every failure gets fixed the same way: add or tighten a text snippet, or adjust the behavior brief. Re-test until 15 answers in a row are correct and useful.

Step 5: Brand it and go live (about 30 minutes)

In the Display settings, match the bot to the client’s brand: their colors, logo as the avatar, a name like “SmileCare Assistant,” and a welcome message that sets expectations (“Ask me about services, prices, and booking — a human reads everything I can’t answer”). Then copy the embed snippet into the client’s website (a Custom HTML block in WordPress, or their site builder’s embed option) and check it on mobile — most local customers will meet the bot on a phone.

Pricing your AI chatbot service

Position pricing as examples, not industry standards — every market differs. A common freelance structure is:

  • Setup fee: typically $500–$1,500 for a first build (data collection, training, behavior brief, testing, install). Charge more when the client has messy or missing documentation, because Step 1 takes longer.
  • Monthly retainer: typically $99–$299/month for reviewing conversations weekly, patching knowledge gaps, updating hours/prices, and sending a short monthly report (“your bot answered 214 chats and captured 31 leads”).
  • Who pays for the builder subscription? Two clean models: the client owns the Chatbase/Tidio account and pays the tool directly (simplest), or you hold the account and bundle it into the retainer with a markup.

Anchor the price to one booked job: if the plumber’s average emergency call is worth several hundred dollars, a bot that captures one extra after-hours lead per month has already paid for itself. Say that sentence on the sales call.

Landing your first clients (without a portfolio)

  1. Build a free demo bot for one business — pick a friendly local business, train it on their public website in an afternoon, and record a 2-minute screen recording of it answering their real FAQs. Send it with one line: “I built this for your site in an afternoon — want it live?”
  2. Fish where owners gather — local Facebook groups, your city’s chamber of commerce directory, and service directories full of businesses with outdated websites.
  3. Sell the after-hours angle — your pitch is one sentence: “You lose every customer who has a question after 6 PM; this answers them and books the job while you sleep.”

FAQ

Do I need coding skills to offer this?

No. Chatbase, Tidio, and custom GPTs are all no-code. The only “technical” step is pasting an embed snippet into the client’s website — if you can add a YouTube video to a page, you can do this.

What if the bot gives a wrong answer?

That’s what the behavior brief and weekly reviews are for. Instruct the bot to admit what it doesn’t know and hand off to a human, then review the conversation log each week and patch gaps with text snippets. Your retainer covers exactly this work.

Should the client or I pay for the chatbot builder?

Either works — just pick one model and be consistent. Client-owned accounts are simpler and build trust; you-owned accounts bundled into the retainer earn you margin. Never let the subscription lapse silently: a dead bot is worse than no bot.

How much monthly maintenance does a chatbot really need?

Plan on 30–60 minutes per client per month: skim the conversation log, add snippets for new questions, update any changed prices or hours, and email a one-page report. That’s the service the retainer pays for.

Watch

Watch this step-by-step walkthrough from Chatbase’s official channel — it covers creating and training a chatbot, customizing its appearance, testing it, and integrating it into a client’s website:

Keep Reading

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top