What is a knowledge base?
Updated · By Robert Breen
A knowledge base is an organized collection of a business's reference information, such as services, prices, policies and how-tos, kept in one place so people and AI tools look up the same correct answers. For AI, it is the material you want answers drawn from instead of guesses.
Why it matters for a small business
Every AI writing tool has the same weak spot: it doesn't know your prices, service area or rules unless you supply them. A knowledge base is the fix that scales. Instead of each person typing facts from memory, the AI reads the current, approved version.
Building one also forces a useful clean-up. Writing down what you actually charge, what's included and what you never promise tends to surface disagreements between staff. AI output can only be as consistent as its source, so a short, current, well-organized knowledge base beats a large, stale one.
In a real lesson: ChatGPT for Electricians: Build an Electrical Bid Writer
In the electrical bid writer lesson, the knowledge base is a single document. You build Electrical Bid Writer for Oak & Wire Electric, a made-up residential and light-commercial contractor in Lancaster County, PA, then attach Oak-and-Wire-Services-Pricing.pdf. It holds the bid-writing rules, the words the company never uses, price ranges for every service and a sample bid.
The instructions tie the GPT to it: use the price ranges in the services guide, never quote a single fixed price before a site visit, and only describe services the company offers (panel upgrades, EV charger installs, whole-home surge protection, GFCI/AFCI updates, lighting retrofits and troubleshooting). Ask for Write a bid description for a 200-amp panel upgrade and the price should come back as a range from the guide.
That one PDF is a small knowledge base for one assistant. Kept current in a shared folder, the same content could also feed a website FAQ or an n8n agent's system message.
Try this lesson free or read the step-by-step guide.
Common confusions
Knowledge base vs training data
Training data shaped the model before you ever used it. A knowledge base is your material, given to the AI when it answers, and you can update it any day.
Knowledge base vs RAG
RAG is a technique for searching a knowledge base and inserting the relevant parts into the prompt. The knowledge base is the content; RAG is one way to use it.
Knowledge base vs system prompt
Short, stable facts (three service areas, two prices) fit in a system prompt, as in the email responder lesson. Longer material belongs in documents the AI can search.
Tips
- Give each document one topic and a clear title.
- Date your price lists and remove old versions.
- Include rules for missing information, like "price confirmed after a site visit".
Related terms
Where you use it: free lessons
- ChatGPT for Electricians: Build an Electrical Bid Writer (ChatGPT, 8 min)
- Build a Custom GPT Plumbing Quote Writer (ChatGPT, 8 min)
- AI Email Responder: Draft Gmail Replies Automatically with n8n (n8n, 12 min)
Prompt templates that use it
Visual guide
Knowledge base in a few slides, with the same guide written out as text.
Frequently asked questions
- Do I need special software for a knowledge base?
- Not to start. A few well-organized documents can serve as one, uploaded to a Custom GPT or kept in a shared folder. Dedicated tools help once the volume grows.
- How do I keep an AI from answering outside the knowledge base?
- Tell it to use only the documents for facts and what to say when something isn't covered, then test it with questions the documents don't answer.