UTIO

AI glossary · Data and knowledge

Knowledge base

In one sentence

A knowledge base is a business's organised store of information (FAQs, terms, opening hours, product sheets, procedures) that an AI agent or a team draws on for answers.

Also known as: KB · help centre · knowledge repository · agent documentation

What is a knowledge base?

Before AI, a knowledge base was the internal wiki or the help centre on the website: the place where answers were written down. With AI agents it has taken on a new role. It’s the source the agent draws on for what it tells your customers.

Put it to work

An AI agent on your WhatsApp, CRM included

From €49/mo + VAT

It replies to customers on WhatsApp and logs every conversation in the CRM.

See the WhatsApp agent

A language model knows a lot about everything, but nothing about your business. It doesn’t know your delivery time to the Channel Islands or whether you open on bank holiday Monday. The knowledge base gives it that information. And what isn’t there, the agent shouldn’t make up.

Related to this term

UTIO WhatsApp

  • 3 users and 500 AI conversations
  • CRM and shared inbox included
  • Built from your website in a minute
  • 14-day free trial, no card needed

From €49/mo + VAT

See the WhatsApp agent

How it works

  1. Gather. Collect what customers ask about: website, FAQs, terms, opening hours, product sheets.
  2. Organise. One idea per section, with no contradictions between documents. If two PDFs give different lead times, the agent may quote the wrong one.
  3. Index. The system splits the documents into chunks and prepares them for search by meaning, usually with embeddings.
  4. Look up. For each question, the agent finds the useful passages and answers from them. That’s the RAG technique.
  5. Maintain. Review the questions that went unanswered and fill the gaps.

Anchoring answers to your documents in this way is called grounding. It’s the difference between an assistant that sounds right and one that says what your business would say.

Example

A phone repair shop in Dublin handles customers on WhatsApp. The same questions keep coming: how long a screen replacement takes, whether they have the battery for a given model, whether there’s a guarantee. The shop puts together one document with turnaround times by repair type, the guarantee, opening hours and how to post a phone in.

The agent answers those questions at any hour. When someone asks about a model that isn’t on the list, it doesn’t quote a price: it notes the model and the contact details and alerts the shop. At the end of the month, the shop adds the models people asked about most.

Knowledge bases at UTIO

At UTIO, the knowledge base is the documentation you load, and in the support vertical it’s the only source of answers: anything not in it doesn’t get answered. The agent takes a note and opens a ticket in the CRM with the details already captured, so your team picks it up with the groundwork done.

To get started, Creator builds the agent by reading your website, in one minute. It works with the WhatsApp agent and the voice agent, and the agent’s memory and the CRM are stored in Spain.

Frequently asked questions

What format should a knowledge base be in?

Clear text, organised by topic, one idea per section. Web pages, PDFs, FAQs or text documents all work. Scanned documents are turned into text with OCR first. The format matters less than the information being correct and consistent.

How often should it be updated?

Whenever something customers ask about changes: prices, opening hours, terms, the catalogue. A good habit is to go through the questions the agent couldn't answer each month and add what's missing.

Can the agent answer something that isn't in the knowledge base?

That depends on how it's set up. For customer service, the safe approach is that it doesn't: if the answer isn't in your documentation, it says so, takes a note and passes the case to a person.

Related terms

  • RAG

    RAG (retrieval-augmented generation) is a technique in which a language model first retrieves the relevant passages from your own documents and then writes its answer from them, instead of relying on what it memorised in training.

  • Grounding

    Grounding is the practice of making an AI model base its answers on specific, verifiable sources, such as your documents, a database or a search, rather than on what it remembers from training.

  • Embeddings

    An embedding is a list of numbers that represents the meaning of a piece of text, an image or audio. Items with similar meaning have embeddings that sit close together, which lets you search by meaning rather than by exact words.

  • AI hallucination

    An AI hallucination is a response from an AI model that sounds convincing but is false or not backed by any source: a figure, quote, price or reference the model invents while generating the most probable text.

  • Human handoff

    Human handoff is the point at which a chatbot or AI agent passes the conversation to a member of staff, together with the context it has gathered, because the case needs it, a rule requires it or the customer asks for it.

Updated on

← Back to the glossary

UTIO is the AI agent line of Universal Telecom, an operator registered with the CNMC.