An AI engine that drafts and schedules social, on voice.
Generates posts, member stories, replies, and video scripts with an LLM, then publishes to X on a schedule with analytics and a voice-QA gate.
Context
An organization producing constant social content needed volume without losing its voice, and without a person hand-writing, scheduling, and posting every item across a busy calendar.
What we built
We built an AI social-content platform. It generates six kinds of output (posts, member stories, replies, video scripts, image design briefs, and translations) via a frontier LLM, runs them through a voice-QA check so they sound like the organization, then publishes to X through OAuth. A scheduler (queue, calendar, cron) handles timing and an analytics sync closes the loop, with full OAuth auth, callback, and publish, plus cron publish and analytics jobs on a Prisma-backed store.
The value
- Turns a person-bound bottleneck into a pipeline. The engine drafts and checks voice, a person approves and schedules each post, and it goes out on time without anyone hand-writing or hand-posting it.
- The voice-QA gate keeps AI volume from sounding like AI. Content scales without going generic.
- Scheduling and analytics close the loop, so posting is a managed calendar rather than a daily scramble.
- Six content types from one engine (posts, stories, replies, scripts, image briefs, translations) feeding the whole social operation.
one generation engine
on-brand gate before publish
a person clears every post before it's scheduled
approved posts publish to X on their own
Client name and identifying details withheld by design. Every figure here is drawn from the system we actually delivered.
Your work would get the same discretion.
Tell us what's breaking. We'll tell you whether we've solved something like it. And if we take it on, your name stays yours.
Start a Diagnostic