Drafts that sound like you, per network, from the start.

Every network has its own physics. X rewards a cold open, LinkedIn hides everything after three lines, Instagram cuts at 125 characters, TikTok wants the payoff in the first second. A single caption sprayed across all of them underperforms on all of them.

Gongong AI writes per platform from the start — not one caption resized — respecting length, tone, hashtag convention and emoji habit. The character counters, the truncation previews and the platform preview renderers all read from one shared table, which is why the counter and the preview agree with each other and with the feed. When we show you the cut point, it is the real one.

Brand voice is a profile, not a prompt suffix. Paste five to twenty of your own posts and extraction captures tone, vocabulary, banned words, sentence rhythm and length, emoji and hashtag habits, opening and closing habits, punctuation and formality. That profile is then applied in five places — drafting, refinement, repurposing, image generation and inbox replies — so the account sounds like itself wherever the words come from. Unlimited voices per workspace, with a default, because an agency needs one per client.

Refinement is an edit rather than a re-roll. The conversation history goes back to the model, the platform limits are enforced, and it is metered as one adjustment — so asking for “punchier” does not cost you the draft you already liked.