---
title: "Knowledge Base Generation"
canonical: https://harmony.io/platform/knowledge-base-generation
---

# Knowledge Base Generation

Automatically generate, maintain, and optimize your knowledge base. Turn resolved tickets, internal documents, and company policies into structured, searchable articles - without asking anyone to write them by hand.

## Knowledge Base Generation

Automatically generate, maintain, and optimize your [knowledge base](https://docs.harmony.io/settings/managing-knowledge-base). Turn resolved tickets, internal documents, and company policies into structured, searchable articles, without asking anyone to write them by hand.

## What is AI knowledge base generation?

Knowledge base generation is producing and maintaining support documentation from work that has already happened, instead of asking people to write it. Every resolved request is evidence: it shows what employees ask, in their own words, and what actually fixed it. Harmony reads that evidence and turns it into structured, searchable articles your team and your AI agent can both use.

It works in two directions. Existing documentation in Confluence, Notion, SharePoint, Google Drive, ServiceNow and Freshservice knowledge bases, Simpplr, Workvivo and uploaded files is indexed into one searchable layer, so employees get answers without knowing where a page lives. Harmony Insights then analyzes resolved conversations and ticket history to find the topics nothing covers, and drafts the missing article with a title, an outline built from the resolutions that worked, the source conversations and how often the question came up.

An empty or stale knowledge base is the main reason self-service fails, which makes this the lever on autonomous resolution: every published article widens what the [AI Service Desk Agent](https://harmony.io/platform/ai-service-desk-agent) can answer on its own.

## How Knowledge Base Generation helps your teams

Auto-generate KB articles: from resolved requests and agent interactions

Unify scattered sources: transform [Confluence](https://harmony.io/integrations/confluence), [Notion](https://harmony.io/integrations/notion), and [SharePoint](https://harmony.io/integrations/sharepoint) pages, PDFs, and past requests into one structured knowledge base

Internal vs. external knowledge: control which articles are employee-facing and which stay internal to your team, so agents see the full picture while employees only get what's meant for them

Keep documentation current: by detecting outdated content and [suggesting updates](https://docs.harmony.io/dashboard-and-analytics/using-harmony-insights-for-kb-and-service-catalog-recommendation)

Power self-service deflection: surface the right article before a ticket is ever created

## Documentation that writes itself

- **#1**: Eliminates the #1 blocker to self-service: empty or outdated knowledge bases

- **90%**: Cuts KB creation effort by 90%: no more chasing teams across the company for articles

### Always Improving

Improves continuously: based on what employees actually search for and what resolves their issues

### Higher Resolution

Feeds directly into the AI Service Desk Agent: for higher autonomous resolution rates

## Powerful capabilities built for enterprise IT

- AI-driven article generation: from request resolutions, past knowledge, and internal docs
- Automatic content gap detection: identifies topics employees ask about with no coverage
- Staleness detection: rewrite suggestions based on changing processes
- Approval workflows: for IT review before publishing
- Multi-format ingestion: Confluence, SharePoint, Google Docs, PDFs, Slack threads, and more

## How knowledge base generation works

**Trigger:** Employees keep asking the same question, and nothing in your documentation answers it.

1. **Ingest what you already have** - Connect Confluence, Notion, SharePoint, Freshservice and ServiceNow knowledge bases, Simpplr and Workvivo, or upload PDFs, internal docs and exported threads. Content is indexed rather than moved, so each source stays with the team that maintains it.

2. **Find the gaps** - Harmony Insights analyzes resolved conversations from Slack, Teams and the portal, and ticket history from a connected ITSM, filtered by date range and category, to detect recurring patterns and the volume behind each one.

3. **Tell a gap from a service request** - A recurring how-to or troubleshooting question becomes a knowledge gap. A repeated request for something to be done - access, hardware, a license - becomes a service catalog suggestion instead. No manual tagging.

4. **Draft the article** - Each gap comes back as a draft with a recommended title, a content outline built from the resolutions that worked, the source conversations and the number of times it was asked, so reviewers can judge priority at a glance.

5. **Review, publish and keep it honest** - Approval workflows put your team in front of anything employee-facing, internal-only articles stay with agents, and staleness detection proposes rewrites when the underlying process changes.

**Outcome:** A knowledge base that grows out of real requests, and an AI agent that resolves more because of it.

## Mammoth Brands went from ~50 articles to 322

Mammoth Brands’ three-person IT team supports around 600 employees across the US and UK. Its knowledge base grew from roughly 50 articles to 322, time to first response fell from four hours to seven seconds, and employee CSAT settled at 4.7 - two weeks from live proof of concept to full rollout.

> Time to response, we went from four hours to seven seconds because the bot responds immediately.
>
> - Marce Ayala, Senior Manager, IT Infrastructure, Mammoth Brands

## Generated knowledge vs. hand-written documentation

Nobody argues against documentation. It loses to the queue, every week, in every IT team.

|  | Hand-written knowledge base | Harmony knowledge base generation |
| --- | --- | --- |
| What gets written | Whatever someone had time for, usually a topic they already knew | The topics employees actually asked about, ranked by volume |
| Who writes it | Senior engineers, in the gaps between tickets | Drafted from resolved conversations, reviewed by your team before publishing |
| Finding gaps | A quarterly audit, or a complaint | Continuous pattern detection across conversations and ticket history |
| Staying current | Articles rot quietly until someone follows bad instructions | Staleness detection proposes rewrites as processes change |
| Where answers live | Spread across Confluence, Notion, SharePoint and someone’s DMs | Every source indexed and searched together, with up to three articles cited |
| Effect on tickets | Employees never find the article | The agent surfaces the answer before a ticket exists |

## Agents that keep knowledge current:

- [Knowledge Base Auto-Generation](https://harmony.io/agents/knowledge-base-auto-generation)

- [Knowledge Article Recommendation](https://harmony.io/agents/knowledge-article-recommendation)

## Ingest knowledge from the sources your teams already maintain

- Documentation: [Confluence](https://harmony.io/integrations/confluence), [Notion](https://harmony.io/integrations/notion), [Microsoft SharePoint](https://harmony.io/integrations/sharepoint)
- Service desks: [Freshservice](https://harmony.io/integrations/freshservice), [ServiceNow](https://harmony.io/integrations/servicenow)
- Employee platforms: [Workvivo](https://harmony.io/integrations/workvivo), [Simpplr](https://harmony.io/integrations/simpplr)
- Manual upload: PDFs, internal docs, and any file format

## FAQ

### What does Harmony's Knowledge Base offering include?

Harmony’s Knowledge Base combines the best of your existing documentation with AI-powered knowledge creation.

**Unified knowledge:** connect and synchronize content from your existing knowledge sources - such as Confluence, Notion, SharePoint, and Freshservice KB - so Harmony’s AI can instantly search and answer from your organization’s collective knowledge.

**Harmony Insights:** continuously improve your knowledge base by automatically identifying knowledge gaps and generating new article suggestions based on patterns detected across resolved support conversations.

### Which external knowledge sources does Harmony sync with?

Harmony connects to and indexes Confluence, Notion, SharePoint, Freshservice KB, and Simpplr, letting the AI agent draw on this content when answering employee requests without employees needing to know where the answer lives.

### How does Harmony Insights generate new KB articles?

Harmony Insights analyzes resolved support conversations and ITSM ticket history to detect recurring patterns. When it identifies a knowledge gap, it generates or updates a KB article suggestion with a recommended title, content outline, source conversations, and volume count.

### How does Harmony tell a knowledge gap apart from a recurring service request?

Harmony Insights classifies recurring patterns automatically: how-to or troubleshooting questions become a knowledge gap suggestion, while repeated requests for something to be done, like software access or hardware issues, generate a Service Catalog item suggestion instead.   
No manual tagging is required.

### What data can Harmony Insights analyze to find KB gaps?

Harmony Insights can run its analysis on conversation history from Slack, Teams, and portal channels, on ITSM ticket history from a connected platform, or on both together for cross-channel pattern confirmation - filterable by date range and category.

### What are the main benefits of Harmony's Knowledge Base offering?

Harmony's Knowledge Base offering closes documentation gaps automatically: Harmony Insights detects recurring how-to questions and drafts a ready-to-publish KB article, including source conversations and volume counts, so teams don't have to guess what to document next. Syncing Confluence, Notion, SharePoint, ServiceNow KB, Freshservice KB, and Simpplr into one searchable layer also means employees get answers without knowing where the source content lives.

### Who reviews AI-generated articles before employees see them?

You do. Every suggestion arrives as a draft, approval workflows route it to the owner you choose, and nothing becomes employee-facing until it is approved.

Articles can also be kept internal, so your agents see the full picture - including steps you would not publish - while employees only get what is meant for them.

### How does this affect our autonomous resolution rate?

Directly. The knowledge base is what the AI Service Desk Agent searches when it answers without a person, so each gap closed widens autonomous coverage, and every response cites up to three ranked articles so the employee can verify it.

Mammoth Brands grew from about 50 articles to 322 and cut time to first response from four hours to seven seconds.