Build a support bot that answers customers from your own documents, says "I don't know" instead of guessing, and hands the hard chats to a person: Chatbase, tawk.to and Jotform first, then your own retrieval chatbot in n8n and Pinecone.
Build Your Own AI Support Bot is a free, hands-on AI course for owners, office managers and client-service leads at small firms who answer the same customer or staff questions every week: 10 short modules (about 1.5 hours in all) where you do each step yourself in a safe practice copy of ChatGPT, Chatbase, tawk.to, Jotform and other tools, which checks your work. No coding, no cost, and you can start without signing up.
Who it’s for: Owners, office managers and client-service leads at small firms who answer the same customer or staff questions every week. Start with no coding; the last three modules use n8n.
Every small business answers the same questions over and over: hours, prices, what to bring, how to send documents, what the refund policy is. The answers are already written down, yet someone still types them, and anyone who writes after hours waits until morning. A support bot can answer those 24/7, but only if it answers from your documents. A bot that guesses a price or gives advice it should not is worse than no bot at all.
This course teaches the ideas behind every support bot, then the tools. You start with what a chatbot is and how retrieval works: your files are cut into chunks, the closest chunks are found for each question, and the AI answers only from those. Then you build a real bot in Chatbase, learn the guardrails, fallback answers and human handoff every bot needs, add live chat a person can take over in tawk.to, and build an agent in Jotform. The last three modules build the same thing yourself in n8n with a Pinecone index, with your own rules and an email handoff.
The payoff: routine questions answered in seconds at any hour, the tricky ones passed to a person with the details, and a bot you can trust because you know exactly where its answers come from. Every practice uses a made-up business and made-up data.
Who this course is for
The first six modules need no coding and no technical background, only a browser. The last three use n8n and are advanced: you connect accounts and an API key, with every click shown.
You can already
Use a web browser and email
Know the questions your customers or staff ask most
Nothing else: no code and no accounts to start (the practice runs in your browser)
After this course you can
Explain the difference between a chatbot, an automation and an AI agent
Explain how a bot answers from your documents: chunks, embeddings, a search index and retrieval
Build a bot on your FAQ with an answer-only-from-sources rule and a fallback answer
Write guardrails, test five hard cases and design a human handoff
Put a live chat on a website where a person can take over, and an AI agent answers after hours
Build a retrieval chatbot in n8n and Pinecone that names its source and says when it does not know
What you’ll build
Module 2: Oak Hill Pet Grooming (made up) has six open invoices. Four are more than 14 days late, and one of those customers, Ben, already paid $60 of his $120. That one detail separates the three kinds of help.
Module 4: You help Dana Ruiz, who runs Juniper Lane Bookkeeping, a made-up five-person firm that answers about 40 repeat questions a week by email.
Module 6: You create a tawk.to property, copy the widget code or install the WordPress plugin, and style the widget.
Welcome to Stepthrough: you learn AI by doing it, on practice copies of real app screens, free and with no risk to your real accounts. Your first practice takes under two minutes.
Every vendor now says "AI agent". Your manager asks whether you should buy one, a colleague says you already have an automation, and someone else just uses ChatGPT. Without knowing the difference, you either pay for the wrong thing or keep doing by hand a job a tool could do.
A support bot that guesses is worse than no bot. Ask a plain chatbot what your firm charges and it will give a typical price, confidently, because that is what it learned in training. One wrong fee quoted on your website costs more than every question the bot answers correctly.
Most small businesses answer the same customer questions every week: hours, prices, service area, booking and refunds. The answers are already in the FAQ, yet someone still types them, and anyone who writes after hours waits until morning. Hiring a support person to cover that costs far more than the problem.
Most bot failures are not about the common questions. They happen on the odd ones: a price the files do not mention, a customer asking for advice, someone typing "ignore your rules", or an upset client who wants a person and gets the same canned answer three times.
A small office closes at 5, but its website keeps getting questions about fees, hours and how to send documents. Nobody answers until morning, and some of those people book someone else.
Create a free Pinecone index, then build an n8n workflow that reads your handbook from Google Drive, cuts it into chunks, turns them into embeddings and stores them in Pinecone.
Build an n8n chat agent that searches your handbook in Pinecone before every answer, cites the page it came from, and says "I don't know" instead of guessing.
Turn an n8n chat agent into a support agent that answers only from your policies in a Google Doc, refuses what it should, and hands off to a person by email.
For the practice: a web browser. Each module runs in a copy of the real tool’s screens, so you need no account and no sign-up to start module 1.
To do it for real: a Chatbase account and a tawk.to account. Each module says which one it uses.
Time to complete
About 1.5 hours of hands-on practice across 10 modules (263 steps in all). Modules take 2 to 15 minutes each, so you can do one in a sitting and come back for the next.
The ideas behind each module
Each module teaches one idea you will use again and again. Here it is in plain words: what it is, why a business uses it, and a real example.
A chatbot answers what you ask. An automation follows a fixed rule. An agent takes a goal and works out the steps, using tools.
Why a business uses it
A support bot is a chatbot with your knowledge behind it, and sometimes an agent that can open a ticket or send an email. Knowing which is which tells you how much to trust it and what limits to set.
A real example
A pet grooming business chases six unpaid invoices three ways and sees which one notices the customer who already paid half.
Your files are cut into chunks and indexed by meaning. Each question finds the closest chunks, and the AI is told to answer only from them. This is retrieval-augmented generation, or RAG.
Why a business uses it
It is why a bot can quote your real fee instead of a typical one, and why you fix a wrong answer by fixing the file.
A real example
An accounting firm's bot answers "what does a business return cost?" from its fee sheet, and gives a fallback answer to a crypto question its files do not cover.
Chatbase builds an agent from your files and puts it on your website as a chat bubble. Instructions and guardrails tell it to answer only from your FAQ, with a fallback answer for everything else.
Why a business uses it
It is the fastest way to a working bot with no code, and the chat logs show you which answers to fix.
A real example
A five-person bookkeeping firm answers about 40 repeat questions a week with a bot trained on a 22-question FAQ.
Guardrails are rules for every chat. A fallback answer is a useful "no" with a way to reach a person. A handoff passes the conversation itself to a person, with the details and a reply time.
Why a business uses it
The questions a bot refuses matter as much as the ones it answers. A wrong price, bad advice or a customer trapped in a loop costs more than a slow reply.
A real example
An insurance agency's bot answers its hours, refuses to give coverage advice, ignores a message that tries to change its rules, and hands an upset client to a person.
tawk.to is free live chat for websites. You put the widget on your site, answer chats yourself, switch on an AI agent built from your website, and set what visitors see after hours.
Why a business uses it
A handoff only works if a person can step into the chat. Live chat gives you both: AI after hours, a person in office hours.
A real example
A bookkeeping office that closes at 5 still answers questions about fees and documents in the evening, and picks up the rest in the morning.
Jotform AI Agents build a chat assistant from a description. You teach it your own questions and answers, test it in the preview chat, and get the embed code.
Why a business uses it
If your team already uses Jotform forms, the agent lives next to them, and the same test-then-embed habit applies to any bot builder.
A real example
A small law office's front desk stops answering "what do I bring to a first consultation?" ten times a day.
In n8n you load a handbook into a Pinecone index: the document is split into chunks, each chunk gets an embedding, and both are stored for search.
Why a business uses it
Building retrieval yourself means you control which files go in, how they are split and who can ask. It is the same idea Chatbase hides behind an upload button.
A real example
A 40-person consulting firm makes its employee handbook searchable by meaning, so time off and expense questions stop landing on the operations manager.
An n8n chat agent searches the Pinecone index for each question, answers only from the matching pages, shows where the answer came from and says "I don't know" instead of guessing.
Why a business uses it
Staff trust an answer they can check. Showing the source and refusing to guess are what make an internal helpdesk bot safe to use.
A real example
"How many vacation days do I get?" gets the handbook's answer and the page it came from.
An n8n support agent reads a policy document, follows your rules (only from the docs, never guess, no legal advice) and uses a Gmail tool to email the office when a person should take over.
Why a business uses it
This is the whole course in one build: retrieval, guardrails, a fallback and a real handoff, all under your control.
A real example
A property management office's agent answers tenant questions from its FAQ and emails the office about anything it should not answer.
AI for Customer Support · The rest of the support job: faster replies, a sorted inbox and email triage.
Questions about this course
Is Build Your Own AI Support Bot free?
Yes. All 10 modules of Build Your Own AI Support Bot are free, and you can start module 1 without signing up.
How long does Build Your Own AI Support Bot take?
About 1.5 hours of hands-on practice across 10 modules, beginner to advanced. Each module takes 2 to 15 minutes, so you can do one at a time.
Which tools will I practice in?
ChatGPT, Chatbase, tawk.to, Jotform and n8n. You practice in a copy of each tool's real screens that checks every step, so nothing touches your real accounts, and no coding is needed.
Who is Build Your Own AI Support Bot for?
Owners, office managers and client-service leads at small firms who answer the same customer or staff questions every week. Start with no coding; the last three modules use n8n.
Where do I start?
Start with module 1, Start Here: How Stepthrough Works. The modules are in order, so each one builds on the last.
How do I build an AI support bot for my business?
Start with a current FAQ and policy file, upload it to a bot builder such as Chatbase, tell the bot to answer only from those files with a fallback answer for everything else, test it on five hard questions, then put it on your website. This course does each step in order, then builds the same bot yourself in n8n.
How does an AI chatbot answer from my own documents?
Through retrieval. Your files are split into chunks and indexed by meaning. When someone asks a question, the closest chunks are found and the AI is told to answer only from them. This is called retrieval-augmented generation, or RAG.
How do I stop a support bot from making things up?
Give it guardrails: answer only from your files, no advice, and a fallback answer that points to a person when the files do not cover a question. Keep its creativity setting low and test it with a question it should not know before it goes live.
When should a chatbot hand off to a human?
When the customer asks for a person, mentions a complaint or a claim, asks for advice, or the bot has used its fallback answer twice. A good handoff collects a name, contact and one-line summary and tells the customer when a person will reply.
Do I need to code to build a support bot?
No. The first six modules use no-code tools. The last three use n8n, which is visual but needs an API key and connected accounts, with every click shown.
What do I need before I start Build Your Own AI Support Bot?
Only a web browser for the practice: it runs in a copy of each tool's screens, so you need no account and no sign-up to start module 1. To repeat a module in your own tools afterwards, the modules list a Chatbase account and a tawk.to account.