The Haiilo MCP integration helps employees find trusted company knowledge through the AI platforms they already use. By connecting Haiilo to AI providers such as Microsoft 365 Copilot and Claude, your organization can reduce information silos, surface relevant knowledge faster, and help work flow more easily.
Haiilo remains the trusted source of truth while AI providers access approved content based on existing permissions and governance controls.
What is MCP?
Model Context Protocol (MCP) is an open standard that allows AI providers to connect to external systems and data sources through a standardized interface. MCP acts as a bridge between an AI platform and business applications, providing access to information and actions while respecting authentication, permissions, and security controls.
Why connect Haiilo to your AI platform?
By connecting Haiilo to an MCP-compatible AI platform, you can extend your trusted company knowledge and news into the AI experiences employees already use every day. Instead of navigating multiple systems and searching across disconnected sources, users can ask natural-language questions and receive answers grounded in relevant Haiilo content, helping them find what they need faster and stay focused on their work.
By enabling the Haiilo MCP integration, you can:
- Deliver trusted company knowledge where employees already work
- Reduce time spent searching across disconnected systems
- Connect AI providers to approved and up-to-date content
- Maintain governance and permissions while expanding access to knowledge
- Help employees get answers faster without adding another destination to visit
- Create, update, and publish wiki articles with AI assistance, within your existing permissions
All access remains permission-aware, ensuring users can only retrieve or edit information or content they have permission to access within Haiilo.
Capabilities overview
The Haiilo MCP integration provides capabilities in these primary areas:
1. Content discovery
Allows AI agents to search, retrieve, and analyze content stored within Haiilo. Examples include:
- Finding company policies and documentation
- Accessing recent news and announcements
- Retrieving content for summarization
2. User and context information
Provides information about users and organizational context. Examples include:
- Looking up employees, communities, groups, and pages
- Retrieving profile information
- Identifying the currently authenticated user
- Understanding user-specific permissions and available resources
3. Content management
Enables authorized AI agents to create and update content in Haiilo, including AI Content widgets and wiki articles. Examples include:
-
AI Content widget
- Discovering editable AI Content Widgets and retrieving their current content.
- Updating widget content programmatically, such as publishing AI-generated summaries, announcements, and status updates
-
Creating and publishing wiki articles with Haiilo MCP
- Creating wiki articles as drafts and updating existing wiki articles
- Publishing wiki drafts after your explicit confirmation
Tools overview
You can determine which tools and permissions are available through the Haiilo MCP connection in Administration > Integrations > Haiilo MCP. Available tools include:
| Tool | What you can use it for | Information retrieved or changed | Example questions |
|---|---|---|---|
search_knowledge_base |
Search for information stored on your Haiilo platform. View supported content types below. For example, you can ask about a policy or how an internal process works. |
Excerpt of the content along with links to the original sources. The most relevant results appear first. | "Where can I find our vacation policy?" |
fetch_full_document |
Open and read the full content of an item found with search_knowledge_base. This gives the AI more information when a search result alone is not enough to answer a question. |
The complete content, details, and information about its source. | "Summarize our vacation policy" |
latest_news |
Find recent content and activity on your Haiilo platform. For example, you can ask what is new across your organization or request updates from a specific team. |
Titles, bodies, authors, publication dates, and links to timeline posts, blog articles, multichannel posts, and events. | "Are there any recent company announcements?" |
search_sender_by_name |
Find a user, community, page, or group on your platform. The tool can also find close matches when the name is not exact. | The name, entity ID, type, profile information, and a link to the matching person or resource. |
"Who is Bob Smith?" "Is there a page for employee onboarding?" |
get_current_user |
Identify the person currently using the MCP integration. This helps the AI understand who is making the request and which resources are available to them. | Profile information and Haiilo identifiers for the authenticated person. |
"Who am I signed in as?" "Which resources can I update?" |
find_writable_wiki_apps |
Find the wiki apps where you are allowed to create wiki articles. | The wiki name and ID, plus the name, ID, and type of the page or community to which the app belongs. | "Which wiki apps can I manage?" |
find_related_wiki_articles |
Find published wiki articles related to content you are about to write. This helps the AI spot likely duplicates or conflicting information before it creates a new article. | Matching published wiki articles, evidence snippets showing why each article matched, and a link to each article. Only articles you are allowed to read are returned. | "Is there already an article about how to request a new laptop?" |
get_wiki_article_for_editing |
Open a wiki article to get its current, editable content. Only articles in the default language that consist of a single rich text editor widget are supported. |
The title and current HTML content, the revision number, draft or published state, last modified date, owning wiki app, and lock information. | "Open the onboarding article so we can fix the outdated section." |
find_publishable_wiki_drafts |
Find wiki drafts that you are allowed to publish, including drafts written by other people. | The draft ID and title, the owning sender and wiki app, and a link to the draft. | "Which wiki drafts are waiting to be published?" |
create_wiki_article |
Create a new wiki article in a wiki you are allowed to write to. By default, the article is saved as a draft, so nobody sees it until it is published. For example, you can create a top-level article or add a sub-article below an existing one. |
The created article, including its ID and link, and its publication state (draft or published). | "Create a draft article on the Finance page in the Guidelines wiki that explains our expense process." |
update_wiki_article |
Replace the content of an existing wiki article, and optionally its title. This works on drafts and published articles and does not change the publication state. Editing a published article changes what readers see immediately. |
The updated article and its new revision number. Readers are only notified if |
"Reorganize this article XX into clear sections and add a summary table." |
publish_wiki_article |
Publish a wiki draft so it becomes visible to its audience. Publishing can trigger notifications, so the AI asks for your explicit confirmation immediately before publishing. |
Confirmation that the article is published. | "The draft looks good. Please publish it." |
findWritableAiContentWidgets |
Find the AI Content widgets that you are allowed to edit. | The widget ID, title, and parent page, when available. | "Find the widgets I can update on the HR page." |
getAiContentWidgetContent |
Review the current content of a specific AI Content Widget before making changes. | The widget's current HTML content, details, and metadata. | "What is currently displayed in the KPI widget?" |
updateAiContentWidget |
Replace the content of an AI Content Widget that you are allowed to edit. The new content is published directly as HTML. For example, the AI can publish a summary, announcement, demo text, or updated status information. |
Updates the widget and returns confirmation that the content was published. | "Publish this XX in the KPI widget: ...." |
Supported content types
| Content type | Content Included |
|---|---|
| Content apps | Title, content within the app, and supported widget content |
| Blog articles | Title, teaser, content within the article, and supported widget content |
| Events | Name, location, start date, end date, description |
| Files | Name, description, file content |
| Forum threads | Thread title and content |
| List entries | List entries and content |
| Pages | Name, description |
| Multichannel posts | Title, teaser, content within the post |
| Timeline items |
Content within the post Note: Our system automatically assigns an AI-generated archiving date to timeline items depending on the post context. Once their content is marked as archived, timeline items will no longer be used as a source in responses. |
| Users | Name, email, all user profile properties |
| Wiki articles | Title, content within the article, and supported widget content |
| Communities | Name, description |
| Documentation | Haiilo's product documentation from https://support.haiilo.com/hc/. Enables users to receive information about how the platform works, such as "How do I create a page?" |
Archived content is excluded. This includes any archived articles, pages, communities, and events.
Content included from widgets
For content apps, blog articles, and wiki articles, the connection can also use content from supported widgets when generating answers.
Supported widgets include Rich Text Editor, Text, Title, HTML, Teaser, My Bookmarks, Callout, Link button, Divider, User profile, and Image (alt text only).
Configuration
You can set up the Haiilo MCP integration for various AI providers. Learn more in:
To use Haiilo MCP, your AI platform needs to support OAuth via PKCE with Authorization Code and a pre-registered Client ID and Secret. Dynamic client registration is not supported.