Connectors
Bridges between the tools your team already uses — Slack, Drive, Confluence, Jira, Notion, Salesforce, HubSpot, Gmail, Loom — and the Agent Workspace.
Connectors are how Headways stays plugged into the rest of your stack. They watch the tools your team already lives in and turn the events that matter into prepared work inside the Agent Workspace.
If your team uses it, a Connector for it is either available or on the roadmap. The current first wave:
- Slack — Mentions, DMs, messages in channels, reactions
- Google Drive — New docs, doc updates, comments, shares
- Confluence — Page created, page updated, comments
- Jira — Issue created, assigned, status change, comment
- Notion — Page created, page updated, database row change
- Salesforce — Deal stage change, opportunity created, custom property change
- HubSpot — Contact created, deal moved, task overdue
- Gmail — New email, label applied, reply received, thread closed
- Loom — Video published, comment added, transcript ready
More sources land as we ship them. See the Connector Catalog for the live list and setup steps.
The Model
Every Connector exposes two things:
- Events — Structured signals from the source tool. ("A message containing 'launch' was posted in #eng-releases." "A deal moved to Closed Won." "A doc was shared with the brand team.")
- Triggers — Rules you author that filter events and decide which Skill runs.
When an event matches a trigger, the workspace fires the named Skill, assembles a draft using the Knowledge Graph, and lands a Shared Conversation in the right teammate's Inbox.
How a Trigger Works
A trigger is a contract between a source event and a Skill:
when {source}.{event} happens
and {filter conditions}
then run {skill}
as {owner}
delivering to {inbox | outbox | both}
For example: when a Slack message in #eng-releases mentions "shipped" and is from a release manager, run the /launch-brief skill as Carl in GTM, deliver to Carl's Inbox.
Skills define what assembly looks like — what context to pull from the Knowledge Graph, what artifacts to draft, what review gates to apply. The trigger just says when to fire.
Why Triggers Matter
Without triggers, AI assistance is reactive: a teammate has to remember to ask. With triggers, the right Skill runs at the right moment without anyone thinking about it. The Q3 launch brief starts drafting itself the moment the eng team posts the release; the renewal-at-risk playbook fires the moment a customer signal lands in CRM.
That's what turns a passive Skill library into an active workspace.
Owner-Direct Writes
Connectors never write back to a source tool on their own. When a Skill drafts a Gmail reply, the reply lands in the Outbox under the owner's authority. It only sends when the owner approves. Same applies to CRM updates, Slack messages, Linear comments, Notion edits, Confluence updates — every outbound write passes through human review.
The owner stays accountable. The agent stays helpful.
Governance and Audit
Connector events, trigger matches, and resulting drafts all flow through the Governance Engine and into the Output Ledger. For any artifact, you can trace:
- Which source event triggered it
- Which Skill ran, at what version
- Which Knowledge Graph entries were used
- Which owner reviewed it, when, and what they changed
- Whether it shipped, was edited, or was discarded
That trail is what makes the platform enterprise-ready.
Related
- Agent Workspace — where Connector-driven work lands
- Shared Skills — what runs when a trigger fires
- Knowledge Graph — what Skills pull for context
- Governance Engine — protection rules around Connector outputs
- Connecting a Data Source — step-by-step setup
- Configuring Triggers — author your first trigger