Zendesk Knowledge Base Automation: What You Can Automate in Guide, and Where AI Helps

This guide walks through what you can set up in Zendesk today, which plan each feature needs, and where AI fits in.

Zendesk features and plan names were checked against Zendesk's documentation in October 2026. Zendesk changes packaging often, so confirm in your own account.

Three jobs worth automating

"Knowledge base automation" covers three different jobs, and Zendesk handles each one with different features:

1. Getting the right article to the right person. Agents should see relevant articles while they work a ticket, and customers should find answers before they write in.
2. Keeping articles correct. Articles need regular review, and shared text should only have to be fixed once.
3. Finding what's missing. You need a steady list of questions the help center doesn't answer yet.

Job 1: Get articles in front of agents and customers

Suggested articles in the ticket

The knowledge section of the Agent Workspace context panel suggests articles that match the ticket's topic and the customer's last reply. Agents can search, link an article to their reply, quote it, or flag it, without leaving the ticket. According to Zendesk's documentation, it's available on Suite Growth and above, or Support with Guide Professional or Enterprise.

Set it up: make sure agents who answer tickets have a role that can use the knowledge section, then encourage linking articles instead of retyping answers. Every link is a small signal of which articles carry the load.

Article recommendations in email replies

For years, the go-to self-service automation was the Autoreply with articles trigger action, which adds up to three recommended articles to the email confirmation customers receive. Zendesk has moved it to legacy, and it stops working by the end of 2026.

What to do instead: if you still use it, plan the move now. Autoreplies are managed under AI > AI agents in Admin Center, and the AI-based versions that predict intent are part of Zendesk's Advanced AI add-on. Either way, recommendations are only as good as the articles behind them, which is where jobs 2 and 3 come in.

Job 2: Keep articles correct

Article verification rules

Article verification lets Guide admins create up to 20 rules that mark articles as unverified on a schedule, for example every product area every six months. The rules run weekly, and Zendesk emails each article's owner a reminder to review and re-verify. See About article verification for details.

The Guide content list brings these together: Needs verification, review status, and content blocks sit in the sidebar.

Set it up:

• Assign an owner to every article. Owners who have left the company don't get reminders, so reassign their articles to a current agent or group.
• Start with one rule for your most-viewed sections, with a short interval, and a second rule for everything else with a longer one.

Where it stops: verification is based on age, not on what changed. An article reviewed last week can still be wrong today if a release changed the steps.

Content blocks

Content blocks are reusable pieces of text, such as a warning, a pricing note or a set of steps, that you insert into many articles. When you update the block, every article that uses it updates too. They're available on Enterprise plans (Zendesk documentation).

Set it up: look for text that appears in more than three articles, like plan limits or a login step, and turn it into a block. The next time it changes, you fix it once.

Team Publishing

On Guide Enterprise, Team Publishing adds review states (In progress, Awaiting review, Ready to publish) and lets you stage changes to a published article without touching the live version. Approved updates can be scheduled to publish automatically.

Set it up: require review for articles in sensitive sections, such as billing, security or legal, and let everything else go straight to publish.

Job 3: Find what's missing

The Search dashboard in Explore

Zendesk Explore includes a Search dashboard for your help center. It shows what people search for, click-through rates, searches with no results, and how many tickets were created. Searches with no results and searches followed by a ticket are your clearest gap list.

Set it up: review the top no-result searches every two weeks. Before writing a new article, retest the search yourself, since some results only miss because of wording, and a synonym or a better title fixes it.

Agents flag gaps from the ticket

When an agent can't find an article, or finds one that's wrong, they can flag it or start a new article right from the knowledge section of the ticket. This works best when it's an expected part of closing a ticket, not an optional extra.

Tag tickets that should become articles

A simple, plan-independent setup uses standard Zendesk features:

1. Create a macro called "Needs KB article" that adds a tag such as kb_needed.
2. Create a view that lists solved tickets with that tag.
3. Have your knowledge owner work through the view every week.

It costs agents one click, and it turns "someone should write this up" into a queue.

Where Zendesk's own automation stops

Job	Zendesk feature	Plan (per Zendesk docs)	Still manual
Suggest articles to agents	Knowledge in the Agent Workspace	Suite Growth+, or Guide Professional+	Linking the article
Suggest articles to customers by email	Autoreplies (legacy trigger ends 2026)	Advanced versions need the Advanced AI add-on	Keeping articles good enough to recommend
Schedule reviews	Article verification rules	Check your plan	The review itself
Fix shared text once	Content blocks	Enterprise	Writing the block
Review before publishing	Team Publishing	Guide Enterprise	Approving
Find gaps	Explore Search dashboard, agent flags, tags	Check your plan	Deciding what to write, and writing it

Look at the last column. Everything Zendesk automates is either delivery or reminders. Writing the new article, or noticing that a published article is now wrong, is still a person's job.

Where AI helps

This is the gap AI tools fill: turning what your agents already wrote in tickets into article drafts. You have three options:

• Zendesk Knowledge Copilot. Zendesk's own AI for knowledge base health, in early access for Suite Professional and above as of October 2026. It recommends articles to create or update based on ticket trends and can draft articles.
• A general assistant like Claude or ChatGPT. Works well for a few articles at a time if you copy tickets in, remove personal data first, and paste results back into Guide.
• A dedicated tool like AutoKnowledgeBase. It checks every solved ticket against your help center and drafts new articles and updates in Guide, so verification and search reports are no longer your only signals.

We compare all of these, with Zendesk's own tools, in how to automatically update help center articles in Zendesk. For the concept behind it, see what an automated knowledge base is.

A one-week setup checklist

• Day 1: Check your plan, and list which of the features above you have.
• Day 2: Assign an owner to every article, and create your first article verification rule.
• Day 3: Create the "Needs KB article" macro, the kb_needed tag and the view.
• Day 4: Open the Explore Search dashboard, and note your top 10 no-result searches.
• Day 5: If you use Autoreply with articles, plan the move before it's switched off.
• Ongoing: Work through the view and the no-result list every week or two, and decide whether AI drafting would save your team that writing time.

If you'd rather have drafts appear for you, AutoKnowledgeBase connects to Zendesk in a few minutes and starts with the next solved ticket. You approve every change. Start a 30-day free trial, or see the getting started guide.