AI for Document360: two ways to add AI to your knowledge base

Rama Adi Nugraha
Written by

Rama Adi Nugraha

Katelin Teen
Reviewed by

Katelin Teen

Last edited September 5, 2026

Expert Verified
Illustration of a Document360 knowledge base with AI writing and answering content

First, a story about a confident wrong answer

I build integrations at eesel, so I spend a lot of my time watching AI read a company's knowledge base and try to answer real questions from it. That job has made me deeply suspicious of demos where the bot always gets it right.

One that stuck with me: a B2B vehicle telematics team, scaling from a couple hundred Zendesk tickets a month toward a few thousand, wired an AI agent to their knowledge base. It started confidently telling customers "yes, we support your car model" for brands that were nowhere in their system. Why? Because a single line in their KB said "we support all models." The AI wasn't broken. It was doing exactly what a knowledge base tells it to do, which is repeat the knowledge base. Their honest summary of the early weeks was "trial and error in the beginning."

That is the whole game with AI on a knowledge base like Document360. The tooling is the easy part. The hard part is knowing what the AI will actually say to a customer before it says it. Keep that in mind as we walk the options, because it is the thing most "add AI to your docs" pitches quietly skip.

What "AI for Document360" actually means

Document360 is a knowledge base platform, not a helpdesk, so "adding AI" splits cleanly into two intents. Getting this split right saves you from buying the wrong thing.

A knowledge base forking into two intents: writing the docs, and answering from the docs
A knowledge base forking into two intents: writing the docs, and answering from the docs

The first intent is authoring: writing, editing, translating, and maintaining the articles themselves. This is where a technical writer or support lead spends their time, and it is where Document360's native AI is strong.

The second intent is answering: taking those finished articles and using them to resolve questions, whether that is a customer on your help site or an agent hunting for an answer mid-ticket. This is the intent that touches your support numbers, and it is the one where you actually have a decision to make, because it is not the only place your customers ask questions. Most of them ask inside your helpdesk, which is a different surface from your docs site. That is the same reason we wrote separate guides for AI for Zendesk, AI for Freshdesk, and AI for Help Scout: the docs are one input, but the queue is where the work lands.

Route A: Document360's native Eddy AI

Everything Document360 does with AI is branded Eddy AI. The important thing to understand up front is that Eddy is not one feature, it is a bundle plus two add-ons, and the line between them is exactly where support teams get surprised.

Document360's Eddy AI settings page showing the AI writer suite, AI search suite, MCP, and a credit counter, as taken from Document360
Document360's Eddy AI settings page showing the AI writer suite, AI search suite, MCP, and a credit counter, as taken from Document360

Bundled into the plan you get the Eddy AI writing agent, AI content and FAQ creation, AI search and answer, auto-generated glossaries, duplicate-content detection, and the MCP server. Sold separately are the AI Chatbot (the embedded widget that answers readers) and the AI Premium Suite (screen capture, step-by-step guides, interactive demos, video recording). Notice which side of that line the customer-facing chatbot sits on.

Writing docs with Eddy AI

For the authoring intent, this is a good fit and I would happily use it. The writing agent lives right in the editor, so you can generate an outline, expand a section, change the tone, split long sentences, and get related-article and tag suggestions without leaving the article you are on. If your goal is to get a backlog of half-written docs into shape, Eddy is doing the job in the place you already work, which is exactly where authoring AI should be.

Answering readers: AI search and the chatbot

For the answering intent inside the docs site, Document360 gives you two surfaces. The first is AI search and answer, which synthesizes a direct answer from your articles instead of returning ten blue links, and cites the articles it pulled from.

Document360's Ask Eddy panel returning a synthesized answer with article citations, as taken from Document360
Document360's Ask Eddy panel returning a synthesized answer with article citations, as taken from Document360

I like that it shows its sources and, refreshingly, prints "AI can make mistakes, please verify important information" right under the answer. That honesty is the correct posture for a KB bot, and it is the flip side of my telematics story above.

The second surface is the Eddy AI Chatbot, the add-on widget you embed on your site. It answers from your content and, when it has nothing, offers to raise a ticket rather than inventing something.

Document360's Eddy AI Chatbot returning "no results found" and offering to raise a ticket, as taken from Document360
Document360's Eddy AI Chatbot returning "no results found" and offering to raise a ticket, as taken from Document360

The chatbot is more flexible than it first looks. In its source settings you can point it at your knowledge base, an external website, raw text, uploaded files, FAQs, and even Zendesk or Freshdesk tickets, all with a content-access configuration and a storage cap.

The Eddy AI Chatbot source configuration showing knowledge base, website, text, FAQ, file, Zendesk, and Freshdesk sources with a storage limit, as taken from Document360
The Eddy AI Chatbot source configuration showing knowledge base, website, text, FAQ, file, Zendesk, and Freshdesk sources with a storage limit, as taken from Document360

Two honest limits to flag, both of which the story from the intro should make you care about. First, the chatbot answers on your docs site, not inside your agents' helpdesk, so it deflects at the help center but does not draft replies in the ticket queue. Second, there is no dry-run over your own historical tickets here: you configure the sources and publish. That is the "flip the switch and see" pattern, and it is precisely how a KB line like "we support all models" turns into a confident wrong answer in production.

The MCP server (the actually modern bit)

The one native feature I would call ahead of the pack is the MCP server. Model Context Protocol is the emerging standard for letting an AI assistant query an external system, and Document360 shipping one means an MCP-aware assistant can read your knowledge base directly. If your team lives in an assistant that speaks MCP, this is a real bridge, and it is rarer than it should be. We walk through what you can and cannot do with it in ChatGPT for Document360, and the same pattern is showing up across wiki tools, which we cover in ChatGPT for Confluence.

What Eddy AI actually costs

Here is where you have to slow down. Document360 no longer publishes flat plan prices. The pricing page is a quote-only configurator headed "customized pricing built around your needs," and it routes you to a sales conversation.

Six inputs feeding a custom Document360 quote: team accounts, workspaces, languages, SSO and security, KB privacy model, and AI Premium Suite usage
Six inputs feeding a custom Document360 quote: team accounts, workspaces, languages, SSO and security, KB privacy model, and AI Premium Suite usage

Document360's own FAQ lists the six factors that move your quote:

FactorWhat it means for AI buyers
Team accountsEditors and reviewers, priced per head
WorkspacesEach separate knowledge base you run
LanguagesLocalization adds to the number
SSO and securityEnterprise auth is a price lever
KB privacy modelPublic, private, or mixed
AI Premium Suite usageThe AI add-on is billed on top

The trial is generous (14 days, no credit card, every feature on), and every quote includes a dedicated success manager and onboarding. But the model means you cannot get an AI price without talking to sales, and the parts most support teams want are add-ons on top of a base you are already negotiating. Users notice. One G2 reviewer's summary of the downside was blunt:

G2

"It is expensive and a lot of new features need a separate license."

And on Reddit, a product manager running it as their KB put it more plainly:

Reddit

"I'm using Document360 as my knowledge base provider and i really don't enjoy them and they are very expensive for what they offer."

To be fair, the core product earns real praise too, holding around 4.5 to 4.7 on G2 across 542 reviews, with the interface and search called out most often. The cost complaint is specifically about the add-on-plus-quote structure, not the software. Our full Document360 pricing breakdown has more, and if you are also weighing a switch, Document360 vs Helpjuice covers that head-to-head.

Route B: layer eesel on top of Document360

If your real goal is the second intent, answering support questions, there is another way to do it that does not depend on Document360's chatbot add-on: point a dedicated AI support teammate at your Document360 docs and let it work inside your helpdesk.

Two routes to add AI to a Document360 knowledge base: native Eddy AI, or the eesel layer on top
Two routes to add AI to a Document360 knowledge base: native Eddy AI, or the eesel layer on top

Full disclosure, this is what I work on, so read this section knowing that. eesel is an AI support teammate that reads your knowledge sources, joins the helpdesk you already run, and resolves tickets. It does not have a one-click native Document360 connector today, so the honest mechanic is: it ingests your Document360 content by crawling the published help center or the URLs you give it, the same way it learns from any docs site. If your Document360 KB is public, that is straightforward; if it is fully private, you would point eesel at the sources it can reach.

Where the two routes differ most:

  • It works in the queue, not just the widget. eesel plugs into Zendesk, Freshdesk, Gorgias, Help Scout and others, so it can draft or auto-send replies where your agents actually work, and pull in Document360 answers as one of its knowledge sources rather than the only one.
  • You can simulate before going live. This is the part that would have saved my telematics customer. eesel can run over thousands of your real past tickets and show you what it would have answered, so you catch the "we support all models" traps before a customer ever sees them, instead of after.
  • Pricing is flat and predictable. No quote, no per-seat fee. You pay per ticket the AI handles.

Here is the eesel cost model at a few team sizes, so you can compare it against a Document360 quote:

Tickets handled per montheesel monthly cost
100$40
500$200
1,000$400
2,500$1,000

One resolved ticket or chat session is one task at 40¢, no matter how many messages it takes, with no platform fee and no minimum (pricing). You only pay for tickets the AI actually handles, so a gradual rollout that routes 200 of your 1,000 monthly tickets to AI costs $80, not the full amount. It is the same model we lay out for AI for HubSpot Service Hub and AI for Salesforce Service Cloud.

So which route should you pick?

There is no single winner here, because the two routes serve the two intents. Match the tool to the job.

If your main goal is...Reach for...Why
Writing and maintaining docsDocument360 Eddy AIThe writing agent lives in the editor where you author
A Q&A widget on your public docs siteDocument360 AI Chatbot (add-on)Native, embeds on the help center, cites sources
Feeding your KB to an MCP assistantDocument360 MCP serverA real, modern bridge for MCP-aware tools
Resolving tickets in your helpdeskeesel layered on topWorks in the queue, simulates on past tickets, flat 40¢/ticket
Predictable, no-sales-call pricingeeselUsage-based, no quote, no per-seat fee

If you only ever needed the authoring half, stop reading and use Eddy AI, it is good at it. The decision only gets interesting once "AI for Document360" really means "answer my customers," because that question does not live inside your docs tool, it lives in your helpdesk. That is also why teams comparing AI for Front or AI for Jira Service Management land in the same spot: the KB is an input, the queue is the job.

Try eesel on your Document360 docs

If the answering intent is what brought you here, eesel is built for exactly that. Point it at your Document360 help center, connect the helpdesk you already run, and it drafts and resolves tickets using your existing docs as knowledge, no rewrite required.

The eesel AI helpdesk dashboard showing ticket activity and AI resolutions
The eesel AI helpdesk dashboard showing ticket activity and AI resolutions

The differentiator I would trust most, having watched confident bots go wrong, is the simulation step: run eesel over your real ticket history and see its answers before a single customer does. Then roll it out gradually, paying a flat 40¢ per ticket it handles. It is free to try, no credit card and no sales call, so you can see how it reads your docs before you commit. Start with eesel.

Frequently Asked Questions

What is Eddy AI in Document360?
Eddy AI is Document360's built-in AI layer. The bundled parts are an AI writing agent, AI search and answer, FAQ and glossary generation, duplicate-content detection, and an MCP server. The AI Chatbot and the AI Premium Suite are separate add-ons on top of that. See the full breakdown in our Document360 AI writeup.
How much does AI for Document360 cost?
Document360 no longer publishes flat tiers. Its pricing page is a quote-only configurator, and AI usage (the AI Premium Suite) is one of six factors that move the price, alongside team accounts, workspaces, languages, SSO, and your knowledge base privacy model. Our Document360 pricing guide walks through what actually changes the number.
Does Document360 have an AI chatbot for customer support?
Yes, but it is a paid add-on rather than part of the base plan. The Eddy AI Chatbot answers from your knowledge base, website, files, FAQs, and even Zendesk or Freshdesk tickets, and it can raise a ticket when it has no answer. If you want an agent that works inside your existing helpdesk instead of only in a docs widget, an AI support teammate like eesel is the other route.
Can I use ChatGPT or Claude with Document360?
Document360 ships a native MCP server, so an assistant that speaks MCP can query your knowledge base directly. We cover the connection options in ChatGPT for Document360. The same MCP idea shows up in wiki tools generally, which we go into in ChatGPT for Confluence.
What is the best AI for a Document360 knowledge base?
It depends on the job. For writing and maintaining docs, Document360's native Eddy AI is the natural fit because it lives in the editor. For answering customer tickets from those docs, a helpdesk-native teammate like eesel that reads your help center, simulates on past tickets, and bills at a flat 40¢ per ticket is usually the more predictable choice. Compare with our other guides like AI for Zendesk and AI for Freshdesk.

Share this article

Rama Adi Nugraha

Article by

Rama Adi Nugraha

Rama is a software engineer at eesel AI with two years of experience writing about B2B SaaS, AI tools, and customer support technology. Based in Bali, Indonesia, he brings a developer's perspective to product comparisons — cutting through marketing copy to what the integrations and APIs actually do.

Related Posts

All posts →
Illustration of a Document360 knowledge base panel with an AI assistant answering a question from the docs
Guides

ChatGPT for Document360: 4 ways to add AI to your knowledge base in 2026

There are four real ways to put ChatGPT-style AI into Document360, from Eddy AI to a build-it-yourself API. Here is what each route does, costs, and skips.

Alicia Kirana UtomoAlicia Kirana UtomoSep 5, 2026
Document360 AI in 2025: A realistic look at its knowledge tools
Guides

Document360 AI in 2025: A realistic look at its knowledge tools

Document360 AI offers helpful knowledge tools in 2025, but can it keep up with modern support needs? Here’s how it compares to connected AI like eesel.

Stevia PutriStevia PutriJul 30, 2025
Illustrated banner showing company documents flowing into a retrieval index that answers a question in plain language
Guides

AI knowledge base: how it works, how to build one, and what breaks

The retrieval stack is mostly solved. What breaks an AI knowledge base is the content behind it, and almost nobody budgets for the cleanup that fixes it.

Alicia Kirana UtomoAlicia Kirana UtomoAug 13, 2026
Illustration of a team reviewing an online knowledge base help center on a shared screen
Guides

The 10 best online knowledge base software tools in 2026

Ten online knowledge base software tools compared on real 2026 prices, the plan tier that actually unlocks the knowledge base, and what the AI on top costs.

Kurnia Kharisma Agung SamiadjieKurnia Kharisma Agung SamiadjieJul 31, 2026
Illustration of a public help center FAQ page with a search box and question cards
Guides

The 10 best FAQ knowledge base software tools in 2026

Ten FAQ knowledge base software tools compared on what actually decides them: whether your answer can reach the four places a customer will ask the question.

Alicia Kirana UtomoAlicia Kirana UtomoJul 31, 2026
Illustration of a support team reviewing help center articles and search analytics
Guides

The 10 best customer support knowledge base software tools in 2026

The average help center completes 10.4% of self-service attempts, not the 30-95% vendors advertise. Here are 10 customer support knowledge base tools, judged on what closes that gap.

Alicia Kirana UtomoAlicia Kirana UtomoJul 31, 2026
Illustration of support tickets flowing into help center articles and back out again
Guides

The 10 best help desk knowledge base software tools in 2026

Ten help desk knowledge base tools compared on the thing that decides them: whether a ticket can become an article, and what the AI reading it costs.

Kurnia Kharisma Agung SamiadjieKurnia Kharisma Agung SamiadjieJul 31, 2026
Illustration of help articles sitting alongside customer records in a CRM
Guides

The 10 best CRM knowledge base software tools in 2026

A buyer's look at 10 CRM knowledge base software tools, with the real 2026 prices, the tier each one hides the knowledge base behind, and what the AI actually meters.

Kurnia Kharisma Agung SamiadjieKurnia Kharisma Agung SamiadjieJul 31, 2026
Illustration of a company wiki with a search bar feeding answers into a team chat window
Guides

The 10 best internal knowledge base software tools in 2026

Ten internal knowledge base tools compared on 2026 pricing, search, and the stale-page mechanic that decides whether your team ever trusts the wiki.

Riellvriany IndriawanRiellvriany IndriawanJul 30, 2026

Ready to hire your AI teammate?

Set up in minutes. No credit card required.

Get started free