What is RAG? Retrieval-augmented generation in plain English
How retrieval-augmented generation lets AI answer questions using your own documents, why it reduces made-up answers, and where businesses use it.
The difference between general AI assistants and a chatbot built on your own content, and which one your customers actually need.
General assistants like ChatGPT are excellent for your own team's research and drafting, but they don't know your prices, policies or services and aren't designed to represent your business to customers. A custom chatbot uses AI models with your own content, answers only within the limits you set, matches your brand, hands over to staff and can connect to your systems.
Many business owners ask, "Why not just use ChatGPT?" It's a fair question. General AI assistants are remarkably capable. But there's a big difference between a tool your team uses and a chatbot that speaks to your customers on your behalf.
Tools such as ChatGPT, Claude, Gemini and Copilot are general-purpose assistants. They're great for:
Many businesses get real value giving staff access to a business plan of one of these tools, with a clear policy on what information can be shared.
They don't know your business. A general assistant doesn't know your current prices, opening hours, delivery areas, policies or stock. It may guess, and guesses presented confidently can mislead customers.
You can't fully control what they say. Customer-facing bots need firm limits: stick to your services, avoid giving legal or medical advice, never promise discounts that don't exist.
They don't represent your brand. Customers expect to talk to your business, in your tone of voice, on your website or WhatsApp.
They don't connect to your systems. A general assistant can't check an order status, book an appointment or create a support ticket in your system without integration work.
A custom chatbot combines a powerful AI language model with your own information and rules:
This approach is called retrieval-augmented generation. Our guide to RAG explains it in plain English.
| General AI assistant | Custom chatbot | |
|---|---|---|
| Knows your business | No, unless you paste information in | Yes, from your approved content |
| Customer-facing | Not designed for it | Built for it |
| Controls and limits | Limited | Defined by you |
| Branding | Provider's interface | Your website, app or WhatsApp |
| Hand-off to staff | No | Yes |
| System integrations | Limited | Bookings, orders, CRM and more |
| Data handling | Depends on plan and settings | Designed around your requirements |
No AI is perfect. A well-built custom chatbot reduces errors by answering from your approved content, saying "I don't know" when it should, and citing sources. It also needs testing before launch and regular review afterwards. Our guide to common chatbot mistakes covers the pitfalls.
When customers chat, they may share personal information. You need to know where conversations are stored, who can see them, how long they're kept and whether the AI provider uses them for training. Business AI services typically offer options that exclude your data from training. Tell users they're talking to an AI and link to your privacy policy.
Custom chatbots can sit on your website, inside a customer portal or app, or on messaging channels such as WhatsApp. A well-designed widget shouldn't slow your site down; our website development team builds them to load without hurting performance.
Our AI chatbot development service builds chatbots trained on your content, with clear limits, staff hand-off and privacy designed in. Our wider AI solutions service can also set up internal assistants and automations for your team.
Custom GPTs live inside ChatGPT and are used by people with ChatGPT accounts. For a branded, customer-facing chatbot on your own site with integrations and controls, you need a custom build.
It's much less likely when it answers from your approved content with strict instructions, but testing and monitoring are still essential.
They can use models from providers such as OpenAI, Anthropic or Google. The choice depends on quality, cost, speed and data requirements.
This guide is general information, not legal, tax or financial advice, and details such as prices, features and rules can change. Product and company names mentioned are trademarks of their respective owners and are used only to identify those products; UK Technology Hub is not affiliated with or endorsed by them.
Tell us about your project and get a free, fixed quote.
Get a free quoteHow retrieval-augmented generation lets AI answer questions using your own documents, why it reduces made-up answers, and where businesses use it.
The most common reasons business chatbots frustrate customers or cause problems, and practical ways to build one people actually find helpful.
How to tell whether an AI chatbot would genuinely help your business, when it wouldn't, and what to prepare before you start.