Glossary

Knowledge base

A knowledge base is the written set of facts a business answers questions from: policies, product guidance and how-to steps. Customers read it in a help centre, your team uses it to reply consistently, and an AI agent searches it to ground its answers. For a Shopify store, it covers what order data cannot: the rules, the reasons and the details.

Last updated: 19 November 2026 · By

Fyncall wrote this page. Claims about Shopify link to Shopify's own help centre, and claims about Fyncall link to our docs. Both were checked 19 November 2026.

What is a knowledge base?

A knowledge base is a maintained library of answers that people or software look up instead of asking someone. In customer service it usually holds your policies, frequently asked questions, product guides and troubleshooting steps.

What is the difference between a help centre, internal knowledge and AI knowledge?

They are three audiences for the same facts. A help centre is written for customers, internal knowledge is written for your team, and AI knowledge sources are the documents an AI agent is allowed to search when it replies.

TypeWho reads itWhat goes in itWhat to watch
Public help centre Customers, before they contact you Shipping, returns, sizing, care, payment and account questions Friendly wording, easy to find, no internal detail
Internal knowledge Your support team Exceptions, escalation contacts, supplier notes, how to handle edge cases Keep it private; it often holds things customers should not see
AI knowledge sources An AI agent, then the customer through its reply The subset of facts you are happy for the AI to repeat Anything in it may be quoted to a customer, so check it first

The last row is the one stores get wrong. An internal note such as “we sometimes refund without a return for loyal customers” is fine for your team. Given to an AI, it can become a promise to every customer who asks.

Where do Shopify store policies fit?

Your Shopify policies are the formal version of your most important answers, so your knowledge base should agree with them word for word on the facts. Shopify lets you manage return, privacy, terms of service, shipping, legal notice, contact and subscription policies, and says they are automatically linked in the footer of your checkout pages once added.

Shopify also says its automated policy templates are for information only, are not legal advice and do not guarantee compliance. The same applies to this page: this is not legal advice. Write the policy first, then write the friendlier knowledge base answer from it.

How do you structure a knowledge base for a Shopify store?

Structure it around the questions customers actually send, with one topic per article or file and the answer in the first lines. Start from last month's conversations, not from a blank outline.

  1. Shipping: regions, delivery times, costs, carriers, split parcels, what to do if tracking stalls.
  2. Returns and exchanges: the window, condition rules, who pays postage, how refunds are issued, exceptions such as sale items.
  3. Products: sizing and fit, materials, ingredients, care, compatibility. Anything a product page leaves open.
  4. Orders and payment: changing an address, cancelling, discount codes, payment methods.
  5. Account and contact: how to reach a person, opening hours, and which channels you answer on.

A few habits make every article more useful:

  • Put the answer first, then the conditions. “You can return unworn items within the window shown in our returns policy” beats a paragraph of background.
  • Use headings, and give each file a name that says what it covers.
  • Do not copy live data into documents. Order status, stock and prices change, so they should come from your store, not from a file written last spring.
  • Give each article an owner, review it when a policy, carrier or product line changes, and delete old versions.

Our return and refund reply templates are a useful starting point for the wording, and canned responses explains how saved replies differ from knowledge articles.

What makes knowledge base content work well for AI?

AI agents answer best from short, specific documents that cover one topic each. Retrieval works by finding the most relevant passages, so a vague or contradictory document leads to a vague or contradictory reply.

  • State numbers and conditions plainly, and avoid “usually” where you mean a rule.
  • Say what is not possible as well as what is, so the AI does not fill the gap with a guess.
  • Test with real customer questions before you let an AI agent reply on its own, and keep a human handoff for anything the documents do not cover.

How does the Fyncall knowledge base work?

Fyncall's knowledge base is a shared library of documents that its AI agents search when they answer customers. The docs describe the search as semantic, not keyword-only: the agent matches the question to passages by meaning and uses the closest matches to ground its answer.

  • You upload PDF, DOC, DOCX, TXT or MD files, up to 10 MB each, and group them into categories.
  • A document is searchable once its status is Ready. You can view the indexed passages, replace a file with a new version, or download the original.
  • The library is central, but each document is linked to an agent in that agent's Knowledge tab, so a WhatsApp agent and a web chat agent can use different sets. See the AI agents overview.
  • “Test Knowledge Base” lets you ask a question and see which sources were matched, with a match percentage for each.
  • Guardrails sit alongside it: the docs put it as “the knowledge base provides the facts, and guardrails enforce the boundaries.”
  • Order, product and customer details come from the Shopify sync, not from your documents.

The AI uses your knowledge base, store data and rules, and it does not retrain itself on your conversations. The Free plan includes 5 knowledge base files and Pro includes unlimited files. See pricing and features.

Common questions

Is a knowledge base the same as an FAQ page?

An FAQ page is one small, public part of a knowledge base. A full knowledge base also holds longer guides, internal notes and the sources an AI searches.

Can I put my existing help centre into Fyncall?

Fyncall's docs describe uploading files, not importing a website, so save your help articles as documents, one topic per file, and upload those.

Does a knowledge base reduce support tickets?

It can, when customers or an AI find the answer before a person needs to step in. That is called ticket deflection. Count the questions on each topic before and after a change, so you know whether it worked for your store.

Ready to try Fyncall?

Install it tonight, let the AI draft your first replies, and switch it on one channel at a time as you start to trust it.

  • Free plan, no credit card
  • 14-day free trial on Pro
  • Unlimited human replies on every plan