The
Engine.
Most "AI social tools" hallucinate features you don't sell, names you've never said, prices that don't exist, and emojis you'd never use.
Foliooo runs a chain of agents that read your brand first, draft second, and grade their own work before you ever see a post.
Voice lock
Three to five real captions from your About page anchor the writer. Drift gets caught before it ships.
Catalog lock
The writer can only reference products and services we actually pulled from your site. No phantom SKUs.
Critic veto
Every draft is read by a second model with the brand's voice profile in hand. Off-brand posts are flagged with the reason.
From URL
to ready-to-ship.
Four agents, in sequence, each with one job. Hand-offs are explicit — the strategist never writes, the writer never plans, the critic never edits. You see what survived the chain, with the reasoning.
Discovery
Reads your URL the way a careful new hire would. Pulls your real products, your tagline, three to five voice samples from your About page, your colour palette, social handles. Caches the brand profile for seven days so a follow-up demo doesn't re-spend the budget.
Strategist
Picks three content pillars from what discovery surfaced, then drafts an angle and hook for each post. No writing yet — just the brief. Operates with memory of your last three weeks of posts so the same angle doesn't ship twice.
Writer
Turns each angle into a caption that reads in your register. Matches the rhythm, punctuation, and tone of your real voice samples. No emojis unless your About page uses them. Also produces a literal image_prompt for the picture pipeline.
Critic
A second model — kept smaller than the writer so it can't just rewrite what it sees — reads every draft with the brand profile in hand and scores 1–5 against your voice samples. A score below 4 fails and is flagged with the reason. You see the score next to every post.
Pictures that
match the words.
Captions get a render, not stock photography. Each post's image_prompt is dispatched to a generation model, polled for completion, and then passed back to a vision Critic that scores the picture on technical quality and brand fit. If the picture doesn't pass, a Director redesigns the scene and we try again. You only see images that survived both rounds.
Vision Critic
A vision model scores the rendered image on craft — composition, lighting, faces, hands, type legibility. Below 4 flags for a regen with corrective notes.
Scene Director
After three failed regens on the same scene, the Director throws the scene out and designs a new one — different setup, same brand. Stops the engine from grinding on a bad idea.
Paste.
Read.
Decide.
The demo runs the same chain a paying customer does — minus the dashboard. Captions land in under a minute. Images render right after. You see what survived.
Try the demo →