Most schedulers know when to post. Reaudit knows what to post about.
A social campaign starts from your AI-visibility data rather than a blank box. Reaudit already tracks which queries answer without mentioning you, which of your pages AI crawlers cite, and which have gone stale. Each of those is a reason to post, and each becomes a draft that says where it came from.
Everything is a draft until you approve it. Nothing reaches a live account without one explicit approval step.
Opportunities
Reaudit lists reasons to post, ranked by priority.
Drafts
Pick up to 10. Each becomes one draft per target platform.
Review
Rewrite, add an image, redesign or drop any single draft.
Approve
One approval schedules the whole set, spaced apart.
In the app this is Dashboard → Social Media → Campaigns. Over MCP or with the in-app AI agent it is the same four steps, as four tool calls.
Four sources feed the opportunity list. Each carries a why and an angle, which is what the copywriter is actually given — so a draft argues a point rather than describing a topic.
Visibility gaps
Queries where AI engines answer without ever mentioning you. The draft names the engines and the query it came from.
Pages that work
Pages AI crawlers already read and cite. Posting about what is already working compounds it.
Content audit
Pages flagged as stale or hard for a model to extract. A post is often the cheapest way to refresh a signal.
Articles
Articles you have already published, turned into platform-native posts.
Grounding is everything Reaudit knows about your business, and it is what stops drafts reading like generic marketing. Copy is written from your ideal customer profile, personas, target market, go-to-market strategy and approved knowledge base.
The Grounding panel on the Campaigns tab shows what is present and what is missing. Thin grounding is the single most common cause of bland drafts.
Crawling produces accurate but generic grounding. The interview fills the gap crawling cannot: why customers actually choose you, which objections keep coming up, what you refuse to do. It asks one question at a time and writes the answers up as a Knowledge Studio document.
/dashboard/assistant?skill=grounding-interviewAlso reachable from the Grounding panel and from Knowledge Studio. The result is saved as a draft document tagged agent://interview — only approved documents enter retrieval, so nothing captured in an interview becomes grounding without your sign-off. Review, edit or delete it in Knowledge Studio before approving.
Select up to 10 opportunities at a time, choose your channels, and click Create. Each opportunity produces one draft per target platform, written to that platform's length and conventions.
The same thing over MCP or the in-app agent:
list_content_opportunities(projectId: "…")
create_social_campaign(
opportunityIds: ["…", "…"],
platforms: ["linkedin", "facebook"],
tone: "professional"
)What should we post about this week? Draft the top two for LinkedIn.
Found 5 opportunities. The strongest are two queries where no engine mentions you. I've drafted both for LinkedIn — text only, no images. Want to review them?
There are two moments to decide, and they give you different options.
| When | Web app | MCP / agent | You get |
|---|---|---|---|
| Before creating | Turn on the Generate images switch next to the Create button | withImages: true, or just say “with images” | An image for every draft, plus the Slides and Headline in image options |
| After creating | Add image or Design one on any draft card | generate_social_image | One image for that one draft |
Images are generated at your project's default quality, in each platform's native frame. Instagram's 4:5 is produced by generating 2:3 and centre-cropping, and the prompt is told about the crop so the subject survives it.
Before any image is generated, a short visual brief is written from the same grounding the copy uses — so the picture is about your customer, not a stock abstraction. You steer it with art direction in project settings: plain language, up to 600 characters, applied to every image on the project.
Two things are refused automatically: text as the subject (screens of code, logs, spreadsheets — models render unreadable pseudo-glyphs, the clearest tell of a generated image), and the usual glowing-dashboard, circuit-board, floating-hologram clichés.
Set Slides to between 3 and 8 and the campaign plans a sequence instead of a single picture. Each slide gets its own scene brief and is told what its neighbours are, so the set reads as one story.
A full editor for post images: templates, text, shapes, your brand palette, and a Library tab holding every image you have generated, uploaded or saved as a brand reference. Click any of them to place it on the canvas at its true aspect ratio.
Open it from Edit in studio on a draft that already has an image, or Design one on a draft that does not — which opens on templates and your library rather than an empty canvas.
Approval covers a whole campaign, so every draft has its own controls — one weak draft never means starting over.
Regenerating always bypasses the image cache, so you never get handed back the exact image you just rejected. Each draft also shows what it was written from — for a visibility gap, the query and the engines that ignored you.
Approve is the only irreversible control on the screen. It schedules every remaining draft, spaced by the interval you choose (60 minutes by default) so a campaign does not arrive as a burst.
approve_social_campaign(campaignId: "…", spacingMinutes: 60)Publishing runs on a cron worker. If it is briefly offline, posts go out on the next cycle rather than being lost.
Engagement is pulled back from each platform every six hours and shown on the Analytics tab — likes, comments, shares and reach where the platform exposes it. Available over MCP as get_social_analytics.
When a fetch fails it now says why, in plain language, rather than reporting zero. A 401 or 403 almost always means the connection is missing an analytics scope and needs reconnecting.
Text generation is cheap; images are the real cost. Prices per image, by quality:
| Quality | Square | Portrait / landscape | Credits |
|---|---|---|---|
| Low | $0.006 | $0.006 | 1 |
| Medium (default) | $0.053 | $0.041 | 1 |
| High | $0.211 | $0.165 | 4 |
A 5-slide carousel is five images. A 10-draft campaign with images at medium quality is roughly $0.50. This is exactly why images are off by default.
LinkedIn engagement analytics
Unavailable. Reading engagement on member posts requires a scope LinkedIn grants only to approved partners. Publishing and scheduling work normally.
Instagram reach and views
Need a reconnect to pick up the insights permission. Likes and comments work without it.
Google Business Profile
Connectable, but posting requires the Business Profile API to be approved on your Google Cloud project.
Headline text in images
Off by default and worth proof-reading. Text baked into a published image cannot be corrected afterwards.
On this page