Sources and Assets¶
Sources and Assets are how content enters Mention. A Source is the connection to an external platform; an Asset is a specific document you choose to include.
Sources¶
A Source is a connection to a content platform your team already uses. Mention supports two types of sources:
Integrations¶
Integrations connect to platforms through a standard authorization flow. You sign in to the provider, grant Mention access, and the platform syncs your available items. Supported integrations include:
- Notion
- Confluence
- Google Drive (including Google Meet transcripts)
- Dropbox
- Box
- GitHub
- Slack (conversation threads, plus channel history as Documents)
- Intercom (Help Center articles)
- Microsoft Teams (meeting transcripts and chat threads, plus OneDrive files)
For details on each, see the Integrations section.
Manual sources¶
For one-off items, you can also add uploads — documents you send straight from your computer. Manual sources behave like any other source once added. See Manual Sources.
Assets¶
An Asset is a specific item you select from a Source for Mention to analyze. When a Source syncs, its items appear as available assets. You decide which ones to activate — only activated assets become source material for the articles Mention writes.
Asset types¶
Mention distinguishes three asset types. Different sources produce different types, and a single source can produce more than one (for example, Google Drive produces both Documents and Transcripts).
| Type | What it is | Where it comes from |
|---|---|---|
| Document | Structured text — pages, files, markdown, web content | Notion, Confluence, Google Drive (files), Dropbox, Box, GitHub, OneDrive, Intercom (Help Center articles) |
| Transcript | Meeting transcripts with speaker turns preserved | Google Drive (Google Meet), Microsoft Teams meetings |
| Conversation | Chat threads with speaker turns preserved | Slack threads, Microsoft Teams threads |
Conversation assets have a special lifecycle worth flagging up front: they are captured on demand when any member of the connected workspace uses @Mention capture in a Slack thread (or mentions the Mention bot in a Teams thread), they get an auto-generated title, and they are activated immediately on capture — there is no inactive state or separate activation step for them. See Managing Assets for details.
Scanned or image-only PDFs (those with no embedded text layer) are transcribed automatically during ingestion, so they produce usable Document content just like text-based PDFs.
Activation limits¶
Each source has a maximum number of documents you can activate:
| Asset type | Limit |
|---|---|
| Documents | 2,000 per integration |
| Transcripts and Conversations | No limit — they don't count against the cap |
Newly synced items default to inactive. You choose which ones to activate from the Sources page.
Sync and refresh¶
Sources stay up to date through automatic and manual sync:
- Automatic sync — All integrations sync daily to detect new or updated content.
- Manual sync — You can trigger a sync at any time from the Sources page.
When content changes are detected, affected assets are re-ingested. Articles follow on a nightly sweep: every night Mention rebuilds each article's supporting context and rewrites the article only when that context has meaningfully changed, so expect up to about a day between a source change and the rewritten article.
What happens after activation¶
Once you activate an asset, Mention ingests its content and extracts the facts it contains into your organization's knowledge base. That knowledge base is what every article is written from.
Activation is the input step. Nothing reaches members until an audience owner adds an entry to that audience's table of contents and Mention writes the article. See Curating Contents.
Activating more of the right material is also the fix when an article comes back not ready — Mention says it cannot cover the entry from what it has rather than inventing the answer.