Email is where missions meet the outside world; connected to Gmail, the agents draft with evidence, follow up on schedule, and send only what your approval policy allows.
The last step of most missions is an email someone has to write. Connected to Gmail, that step closes: research becomes outreach drafts with the evidence cited, reports become sends to the people who asked, and follow-up sequences run on their schedule instead of on memory. The drafts read like a person wrote them because a person set the rules they were written under.
Sending is the most audited write in the platform: every outbound message honors the approval policy, the audit log records what went to whom, and your brand voice settings shape every draft. Autonomy over email is earned through exactly this much control.
Outreach written from actual research - the recipient context and the claim sources attached for review.
Follow-ups and nudges sent on cadence by standing missions, with silence respected as an answer when you configure it so.
Briefs and digests mailed to stakeholders when missions finish - the loop closed without a copy-paste.
Missions read relevant threads before advising, so recommendations respect conversations already in flight.
Real missions from the playbook library - run them as-is from the console, or read the full article behind each.
Find companies actively researching AI customer support this week, rank them by intent momentum, send the top 20 with context to #sales in Slack, and email me the full li…
Find people speaking at marketing conferences this quarter whose companies fit a 100-1000 employee B2B profile, get verified emails, save the list to Google Sheets, and e…
Take my ten highest-intent target accounts, research each company's current initiative, generate a personalized landscape image speaking to it - their industry, their mom…
Find 20 companies that just posted SDR openings, draft a first-touch email to each VP Sales referencing the posting and one company-specific insight, and create the draft…
Connections are scoped to one workspace, every card has a disconnect button, and outbound actions honor your approval policy.
Only under a policy you set. Gated is the default posture for teams starting out: every send waits with recipient, subject and body shown. Teams open it once trust is earned - and the audit log records everything either way.
Missions read what their objective requires - the thread with a prospect, the label you point them at. Access is workspace-scoped, actions are logged, and disconnecting revokes it all.
The brand kit and your saved preferences shape every draft, and corrections stick: edit an agent draft harshly once and workspace memory learns the register you actually send in.
One connection, eight specialist agents, one budget ledger - and your approval policy on everything outbound.