Brand & multi-brand studio
Save your brand's look and voice once — colors, fonts, logo and tone — so everything Sosyabot's AI makes for you stays on-brand, and keep each of your brands neatly in its own place.
Who it's for
- Anyone who wants a consistent look and feel across every post, without re-explaining their style each time.
- Solo creators and small teams who want AI designs and captions that already sound like them.
- Agencies and freelancers juggling several clients or brands, each needing its own colors, fonts and voice.
What you can do
- Build a brand from a single image — upload a logo or a past post and Sosyabot suggests your colors, fonts and tone to start from.
- Lock your colors and fonts so AI designs always use your palette, never random ones.
- Save your tone of voice so AI captions sound like your brand, not generic.
- Keep many brands side by side in one workspace, each with its own look and voice.
- Let your AI helpers and automated agents pick up the right brand automatically, so output stays consistent without extra work.
Getting started
Open your brand settings
Go to your workspace and open the brand area. Your existing style is already there as your default brand, so nothing breaks if you've used Sosyabot before.
Add or build a brand
Create a new brand and fill in its colors, fonts, logo and tone — or upload a logo or sample post and let Sosyabot suggest a starting point you can tweak.
Turn on Brand Lock (optional)
Switch on Brand Lock to force every AI design and caption to stick to that brand's colors and fonts. Leave it off if you'd rather let AI suggest variations.
Generate on-brand work
Use AI design generation and captions as usual — pick the brand you want, and the result already matches its look and voice.
TIP
Running an agency? Give each client its own brand. Switching brand changes the whole look and voice in one click — no need to re-enter anything.
For developers
Brand DNA Extractor
Upload a logo, sample post, or screenshot and Sosyabot returns a starter brand kit you can drop straight into a brand.
| Endpoint | POST /editor/brand-kit/extract |
| Body | multipart/form-data with a single file field (image, max 25 MB) |
| Auth | Authenticated + workspace-scoped |
| Returns | { colors, fonts, voice } |
Colors are extracted deterministically: sharp downsamples the image and quantizes pixels to find up to 6 dominant hex colors, with near-duplicate shades collapsed. Fonts and voice are AI-suggested — a Qevron vision pass describes the visual style, then a text model picks a display/body font pairing restricted to the editor's available font set plus a one-line brand-voice descriptor. The color extraction is the core deliverable and always succeeds; the AI suggestions are best-effort and degrade gracefully if AI is unavailable.
Multi-Brand Command Center
Each workspace can hold many named brands. A brand carries colors, fonts, logo, voice, persona, image style, and a Brand Lock flag.
| Method | Endpoint | Purpose |
|---|---|---|
GET | /brands | List all brands in the workspace |
GET | /brands/:id | Fetch one brand |
POST | /brands | Create a brand (requires an active subscription) |
PUT | /brands/:id | Update a brand |
DELETE | /brands/:id | Delete a brand |
Brands are strictly workspace-scoped and never leak across workspaces; brand names are unique per workspace. A brand supports up to 32 colors and 16 fonts. The legacy workspace brand kit remains the implicit Default brand: when no brandId is supplied, AI falls back to it, so existing single-brand setups keep working with no migration.
Brand Lock
When Brand Lock is enabled on a brand, AI design becomes deterministic rather than suggestive. For template and carousel generation, Sosyabot derives a 7-role palette directly from your brand colors and a font pairing from your brand fonts, then hands these to the layout composer — the LLM never chooses colors or fonts. In the in-editor assistant, Brand Lock injects an explicit instruction to use only your brand colors and fonts and never introduce others.
Brand Voice DNA
Every brand has a voice and persona. These are resolved through a single shared resolver (resolveBrandContext) and injected into all AI text generation: template/carousel copy, the in-editor assistant ("match this tone in any copy you write"), and the autonomous agents. Agents bind to a brand via brand_id; their generated captions and Reels inherit that brand's voice and persona automatically. When an agent has no brand_id, it falls back to the workspace default brand.
AI Design Generation
| Endpoint | Output | Cost |
|---|---|---|
POST /editor/ai/generate-template | A single on-brand design draft, opened in the editor | 10 AI credits |
POST /editor/ai/generate-carousel | A cohesive multi-page carousel sharing one palette and voice | 10 AI credits |
POST /editor/ai/agent | One Qevron tool-calling turn over the live canvas | — |
Template and carousel generation accept an optional brandId and apply the brand's colors, fonts, voice, persona, and Brand Lock. Both require an active subscription and the ai.template feature.
The editor AI assistant (/editor/ai/agent) is agentic: it runs one Qevron turn with the client-supplied tools (describe canvas, add/update/delete objects, set background, place images, etc.) and returns the assistant message plus any tool calls. The frontend drives the multi-turn loop and executes each tool call against the canvas. The brand context — including Brand Lock, voice, and persona — is folded into the system prompt for every turn.