Social post drafts

Ask for a post. Get a post.

Not a paragraph of chat text you have to copy somewhere. Ashley hands you a draft card: a caption, hashtags, an image she had generated during the run, and a toggle for each platform. It sits in your conversation, it survives a reload, and every part of it is editable — by clicking, or just by asking.

What's in a draft card

The character count runs against each platform's real limit — X 280, Instagram 2 200, LinkedIn 3 000, Facebook 63 206 — so you find out you're over before you publish, not after.

Ask for one post

1. Open a conversation with Ashley.
2. Describe the post. Product, angle, tone, platform — whatever you have.
3. Watch the steps: she writes the caption, then the image is generated during the same run.
4. Edit anything you don't like — in the card, or by talking to her.
5. When you're happy, tap a platform toggle and publish. See Publishing a post.

A normal request like that gives you exactly one image. Ashley is instructed never to invent a carousel — if you wanted three images, you have to say so.

Ask for several posts at once

Ask for a batch and you get a reviewable stack of drafts instead of one card. Each one is independently editable, and you address them by index.

Then:

Indexing works for image edits too: "on post 3, make the image brighter" targets exactly that one.

Carousels — only when you ask

Say the word and you get a real carousel: 2 to 10 coherent images, not one image repeated.

The variants are genuinely different, by construction. Ashley works from a set of codified shot directives — a tight telephoto shot, a macro close-up, a wide environmental frame, a flat-lay from above, one with a human using the product, one with seasonal staging — and each generated variant is compared against the reference with a perceptual-hash check. If a variant comes back too similar, it's rejected and retried (up to twice). You get four different photographs of the same brand, not four copies.

Edit images by talking

You don't need to open an editor. Three things you can say, and each is a distinct operation:

All three accept a target. "Make image 2 brighter" in a single post; "on post 3, add another image" in a bulk set.

Your logo and your colours

Ashley can pull your real brand kit and put it on the image.

She reaches for your light logo first, falls back to the dark one, and if you haven't uploaded either she asks you rather than inventing something. The logo is overlaid on the generated image, not described in the prompt and hoped for.

Set your logos and colours up first — see Brand kit.

Products and styles from your Neural Core

This is the part that makes the images actually look like your product rather than a stock approximation.

If your prompt names a product you've saved in your Neural Core, Unyo finds it automatically — no special syntax, no tool to invoke — and feeds that product's real saved photos in as visual references, along with any styles you've saved for it. You just write the product's name.

You can also point at a saved image directly:

…or ask her to reproduce a saved style on a new subject. See Products in your Neural Core and Files.

Attach a photo and get a caption about what's in it

Drag a photo into the chat and ask for a post. Ashley analyses it first — description, visual style, mood, key elements — and the caption is written about what is actually visible in the photo, in your configured language. Not a generic caption bolted onto your image.

This is the fastest route from "I took a photo on my phone" to "that's posted."

Your image library

You get an inline gallery of your image library. Point at one and ask for a post from it, and it becomes the post image.

Going the other way, once you like a post's images:

Each image is copied into a colour-coded folder in your knowledge base, with its generation prompt kept as the description — so you can find it again by describing it.

Hashtags that come from your own numbers

Ask for hashtag help and, once you've connected an account and have posts to learn from, Ashley ranks your proven performers by their average likes — tags you've actually used at least twice — rather than inventing a list. She also tells you each platform's ceiling: Instagram 30, Facebook 10, LinkedIn 5, X 3.

With no connected account and no history there's nothing to rank, so she falls back to suggestions for your niche — and she'll tell you which are proven from your own numbers and which are only suggestions. See Social analytics.

Limits worth knowing before you publish

Two more, stated plainly:

• Ashley never publishes or schedules on her own. Not in a conversation, not in an automation. Her automation persona hard-codes it: prepare the draft fully, the user publishes it from the card with one tap. There is no unattended publishing anywhere in Unyo.
• Ashley doesn't have an image-editing toolbox of her own. Image generation belongs to Tyron, and it's blocked for Ashley at the runtime layer, not just in her instructions. Images reach a post through the draft's own image prompt, the talk-to-edit commands above, your library, or your attachment. It's the same principle as Maya being the only agent with your inbox: each agent's boundary is enforced, not merely requested.

Scheduling and publishing

Drafting is this page. The next two steps have their own:

• Publishing a post — one draft, up to four platforms, in parallel.
• Scheduling a post — pick a time from the card, in your own timezone.

Troubleshooting

The image doesn't look like my product. Save the product in your Neural Core with real photos, then name it in your prompt. Without saved photos there's nothing to reference. See Products.

Ashley made a carousel and I only wanted one image. Ask her to remove the extras — "keep only the first image". She's instructed not to invent carousels, so this usually means the request read as a multi-image one.

"Put my logo on it" and she asked me for a logo. You haven't uploaded one. Add a light logo (and ideally a dark fallback) in your Brand kit.

I asked for an exact pixel size and didn't get it. Output dimensions aren't guaranteed to the pixel. Aspect is driven by the prompt. Ask for "square" or "portrait" rather than "1080×1350".

A platform toggle is greyed out or blocked. A platform I want isn't in the list. The card only lists accounts you've already connected. Connect it from Integrations in the sidebar and it appears in the cross-post list.

FAQ