Overview
Connect your tools to capture feedback and take action
Modem connects to the tools where your users and team already work. Integrations serve two purposes: capturing data (messages, issues, emails) so Modem can analyze it, and giving the agent tools to take action on what it finds.
Integrations are grouped by what they bring into Modem. Most are connected from Settings → Integrations in the dashboard. External MCP servers have their own page under Settings → External MCPs, and each member adds their own Cursor or Devin API key under Profile → Coding Integrations.
Chat
The conversations where your team and customers already talk.
Slack
Capture messages from selected channels. Also lets you talk to the Modem Agent directly from Slack.
Discord
Capture messages from Discord servers. Also lets you talk to the Modem Agent by mentioning the bot.
Microsoft Teams
Capture channel messages from Microsoft Teams.
Social
Public conversations about your product.
Customer Support
Tickets, support chats, and customer-reported issues.
Intercom
Capture conversations in real time. The agent can also search, reply, and manage tags.
Zendesk
Capture Zendesk tickets in real time, including replies, internal notes, and ticket metadata.
Pylon
Capture support issues via triggers. The agent can also search, reply, and add notes.
Plain
Capture support threads: customer emails, chats, and internal notes.
Front
Capture shared inbox conversations: customer messages, replies, and internal comments.
Canny
Capture feature requests, bug reports, comments, and votes from your feedback boards.
Route email feedback to Modem through a dedicated inbound address.
In-app Feedback
Collect feedback from your own app, CLI, or AI agent with a widget, an endpoint, or an MCP tool.
Video Conferencing
Recorded meetings and sales calls, captured as searchable transcripts.
Gong
Capture transcripts from your recorded Gong calls, linked to the people who spoke on them.
Fathom
Capture transcripts and summaries from your Fathom-recorded meetings, linked to the people who spoke on them.
Sales & Revenue
CRM records and billing signals from your go-to-market stack.
Salesforce
Query standard and custom CRM objects read-only, with core account and opportunity context kept synced.
Stripe
Sync a read-only mirror of your billing data and trigger automations on subscriptions, trials, and failed payments.
Project Management
Planning systems, so signals can become work.
Linear
Search, create, and update issues. Link topics to existing tickets.
Jira
Capture issues and comments in real time. The agent can also search, create, and update issues.
Jira Service Desk
Capture customer support requests from Jira Service Management, including public replies and internal notes.
Notion
Search, read, and create pages and database entries in your Notion workspace.
Product Analytics
Behavioral context from your analytics tools. Both are query-based: the agent asks them questions on demand, and nothing is ingested into Modem.
PostHog
Query product analytics, feature flags, experiments, and error tracking data.
Mixpanel
Query events, funnels, retention, and segmentation.
Code & Coding Agents
Repository activity, error tracking, and coding agents.
GitHub
Sync issue and PR metadata: titles, descriptions, comments, labels, and review discussions. Does not read source code.
GitLab
Sync merge request, issue, and comment metadata. The agent can also search, create, update, and comment.
Sentry
Ingest user feedback into topics. The agent can also search issues, read stack traces, and resolve or assign them.
Coding Agents
Delegate implementation tasks to Claude Code, Cursor, or Devin, enriched with Modem context.
Extending Modem
Connect your own tool servers so the agent can use tools Modem does not provide natively.
Looking for the opposite direction, using Modem from Claude Code, Cursor, or another MCP client? See the Modem MCP Server.
What Each Data Source Captures
These integrations continuously capture messages and events, feeding them into Modem’s analysis pipeline. Messages are grouped into topics, classified by type (bug, feature request, praise, complaint), and linked to people and companies.
| Source | What’s captured | Historical backfill | Real-time |
|---|---|---|---|
| Slack | Messages, reactions, threads | 90 days | Yes |
| Discord | Messages, reactions, threads, forum posts | 90 days | Yes |
| Microsoft Teams | Channel messages, replies, reactions, edits | 90 days (≤12,500/channel) | Yes |
| GitHub | Issue/PR titles, descriptions, comments, labels | 90 days | Yes (webhooks) |
| GitLab | MR/issue titles, descriptions, comments, labels | 90 days (≤100 comments per item) | Yes (webhooks) |
| Sentry | User Feedback reports | 90 days (≤50,000 reports) | Yes (webhooks) |
| Intercom | Conversations, replies, notes, CSAT, contacts | 90 days (≤5,000 convs) | Yes (webhooks) |
| Zendesk | Tickets, replies, internal notes, ticket metadata | None | Yes (webhooks) |
| Pylon | Issues, customer and teammate replies, internal notes | 90 days (≤5,000 issues) | Yes (webhooks) |
| Plain | Thread emails, chats, internal notes, thread metadata | 90 days (≤10,000 threads) | Yes (webhooks) |
| Front | Conversations, replies, internal comments, status | 90 days (≤2,500 convs) | Yes (webhooks) |
| Canny | Board posts, comments, votes, post status | Full history (≤15,000 posts) | Yes (webhooks) |
| Jira | Issues, comments, issue metadata | None | Yes (webhooks) |
| Jira Service Desk | Requests, public replies, internal notes | 90 days (≤15,000 requests) | Yes (webhooks) |
| Stripe | Customers, subscriptions, invoices, payments | Full history (on connect) | Yes (webhooks) |
| Gong | Call transcripts, participants, call metadata | 90 days (≤2,000 calls) | Polled (~15m) |
| Fathom | Meeting transcripts, summaries, participants | 90 days (≤400 meetings) | Yes (webhooks) |
| X | Public posts matching your queries, authors, replies | None | Polled |
| Email body and threads | None | Yes |
For Pylon, Plain, Canny, and Jira Service Desk, the backfill is optional, and an admin starts it from the integration’s settings page. The other backfills start automatically.
All data sources let you choose what to monitor: specific Slack channels, Discord channels, GitHub repositories, Pylon triggers, X search queries, or a dedicated email address. Start narrow and expand as needed.
Setup
Most integrations are configured from Settings → Integrations in the Modem dashboard. Data sources connect through OAuth (Slack, Discord, Intercom, Zendesk, Front), an app install (GitHub, Sentry), an API key (Pylon, Plain, Canny), or a generated address (Email). Agent tools use OAuth (Linear, Notion, Mixpanel) or API keys (PostHog, Coding Agents). External MCP servers use no auth, a bearer token, custom headers, or OAuth 2.1, depending on the server.
Each integration page has step-by-step setup instructions.
Cross-Platform Identity
When you connect multiple data sources, Modem links people across platforms using shared identifiers like email addresses. A user who messages you on Slack, files a GitHub issue, and emails support gets a single unified profile with their full history.
This also works at the company level. Email domains and other signals help associate people with companies, giving you an account-level view of feedback across all sources.
See Companies & People for details.