Dispatch runs a small newsroom of AI agents — research, fact-check, plan, and draft — for every client account you manage. Nothing goes out until a human approves it.
Six stages, six agents, one queue you control. Every project runs its own version of this pipeline, on its own sources and brand voice.
Scans your approved sources and surfaces only what clears the relevance bar — headline, summary, source tier, suggested angle.
Checks facts, compliance risk, and brand fit — three independent verdicts. All three have to clear before anything moves on.
Reviews last week's engagement, then plans the week's themes from what's verified and what you've flagged as a priority.
Writes the caption and hashtags, plus a structured image brief specific enough for someone else to execute.
A human reviews every draft in the queue. Approve, reject, or send it back — nothing skips this step.
Copy the caption and image, or let Dispatch publish straight to the platform once you've approved it.
Each agent has one job and stays in its lane — behavior comes from your project's briefs, not from custom code per client.
Watches your approved sources and surfaces only what's relevant, with a credibility tier and a suggested angle for every finding.
Fact-checks, checks compliance exposure, and checks brand fit — independently, before anything counts as usable.
Owns the content mix and posting cadence, and reviews last week's results before planning the next one.
Writes the caption and hashtags in your brand voice, plus a structured brief for the image.
Takes in your specials, events, and must-avoid topics, and reports back in plain language on what got posted.
Watches comments and DMs for buying intent, logs the lead, and drafts a reply — for you to send, never itself.
The Sales and Content Creator agents can draft, but they can't publish or send anything on their own. Every caption, every reply, every image sits in a queue until someone on your team approves it.