Creating Content
ViralToby supports two ways to create content: autonomous generation by Toby and manual creation by you. Both methods produce the same high-quality output -- the difference is who initiates the process.
Two Ways to Create
Autonomous (Toby)
When Toby is enabled, it continuously monitors your content buffer and generates new content automatically. Toby selects topics, hooks, personalities, and visual styles based on your Content DNA and its learned strategy scores. All Toby-generated content goes through the Pipeline for your review before publishing.
Manual (Creation Wizard)
You can also create content manually at any time using the Creation page. Manual creation lets you:
- Specify the exact title and content lines
- Choose the content format from all 10 available formats
- Select which brands to generate for
- Pick a specific music track or let ViralToby choose automatically
- Provide your own idea or script in Semi-Manual mode
Manual content follows the same generation pipeline as Toby's content — same quality scoring, same rendering engine, same output quality.
Creation Wizard
The Creation page walks you through a 5-step wizard each time you generate content.
Step 1 — Content Type
Choose what kind of content to create:
- Video — Reels and short-form video (Content A or Content B formats)
- Posts — Static image posts including carousels, slides, and single-image formats
- Thread — Text-only posts for Threads and Bluesky
Step 2 — Brands
Select which brands to generate content for. You can select all brands or pick individual ones. If your account only has one brand, this step is skipped automatically.
Step 3 — Format Config
Choose the specific format within your selected content type. Each format has its own visual style and rendering pipeline. The format selector shows a preview card and description for each option. See the Format Reference below for the full list.
Step 4 — Mode
Choose how AI and your input are combined:
- 100% Automatic — AI handles everything. Toby generates title, content, and visuals from your Content DNA. This is the fastest path and requires your Content DNA to be configured.
- Semi-Manual — You provide the idea or script. AI handles rendering, image generation, and caption writing. What you need to provide varies by format — see Auto vs Semi-Manual below.
Step 5 — Content Count
Choose how many pieces to generate per brand:
- Videos and Posts — 1 to 3 per brand
- Threads — 1 to 10 per brand
Format Reference
ViralToby supports 10 formats across two content categories. Every format uses your brand's Content DNA and design settings.
Video Formats
| Format | AI Images | Description |
|---|---|---|
| Content A | Yes — AI background | Animated text overlay on a full-screen AI-generated background image, with music. The classic short-form viral reel format. |
| Content B | Yes — AI or real image | Ken Burns pan-and-zoom motion from a single AI-generated or sourced image, with narration-style text and music. Great for story-driven content. |
Post Formats
| Format | Type | Slides | AI Images | Description |
|---|---|---|---|---|
| Carousel A | Carousel | 5 | 1 cover | Classic multi-slide carousel with a branded cover image and up to 4 content slides. Includes your brand header. |
| Slide A | Single image | 1 | 1 | Single branded image post with a "Read Caption" call to action. The full content lives in the caption. |
| Carousel B | Carousel | 4–5 | Per-slide | Advanced multi-slide format where each slide gets its own AI-generated background image and key point. Higher visual variety. |
| Slide B | Single image | 1 | 1 | Single image with an AI-generated cover and highlighted key points. Visually similar to a carousel cover used as a standalone post. |
| Slide C | Single image | 1 | None | Flat-color text image using your brand palette. No AI images — fast to generate, clean aesthetic. |
| Carousel C | Carousel | 5 | None | Multi-slide flat-color text carousel. Same aesthetic as Slide C but with multiple slides. |
| Slide D | Single image | 1 | None | Tweet-style post with your avatar and a text quote layout. No AI images. |
| Carousel D | Carousel | 3 | None | Multi-slide tweet-style format. Great for lists and short-form content in a conversational layout. |
Thread Formats
Threads are text-only posts. ViralToby supports 8 thread format types that shape how the content is structured:
| Thread Format | Description |
|---|---|
| Auto Mix | Toby selects the best format based on your Content DNA topic |
| Value List | A numbered list of actionable tips or insights |
| Controversial Take | A bold or counterintuitive position on a common belief |
| Myth Buster | Debunks a widely held misconception in your niche |
| Thread Chain | A connected series of posts that build on each other |
| Question Hook | Opens with a provocative question that drives replies |
| Hot Take | A short, punchy opinion designed for engagement |
| Micro Story | A brief narrative arc with a setup, tension, and resolution |
Auto vs Semi-Manual
Automatic Mode
In automatic mode, the AI generates everything from your Content DNA:
- Title and key points
- Image prompts and AI-generated visuals (for image formats)
- Caption tailored to each platform
- Music selection (for video formats)
Automatic mode requires your Content DNA to be configured. Without it, the AI has no brand context to draw from.
Semi-Manual Mode
In semi-manual mode, you provide the creative direction and the AI handles production. What you need to provide depends on the format:
| Format | What you provide in Semi-Manual |
|---|---|
| Content A | Title + 5–12 content lines (key points) |
| Content B | Title + narrative paragraphs (one per slide) |
| Carousel A | Title + up to 4 key points (one per content slide) |
| Slide A | Title + long-form caption text |
| Carousel B | Title + slide titles + optional image hints |
| Slide B | Title + key highlights |
| Slide C | Title + body text |
| Carousel C | Title + per-slide body text |
| Slide D | Your tweet text |
| Carousel D | Up to 3 tweet-style text blocks |
| Thread | Opening hook + the thread body |
The AI fills in everything else: rendering, image generation, captions, hashtags, and platform-specific formatting.
Multi-Brand Generation
When you select multiple brands in Step 2, ViralToby creates one generation job per brand. The jobs run in parallel — each brand's content is generated simultaneously, not sequentially.
For example, if you select 3 brands and set count to 2, you get 6 jobs running at the same time. Each job appears in the Pipeline as a separate card as it completes.
Quality Tiers
For formats that use AI image generation (Content A, Content B, Carousel B, Slide B), you can choose between two rendering modes:
| Tier | Speed | Image Quality | Best For |
|---|---|---|---|
| Best Quality | Slower | Highest resolution and detail | Final content, published posts |
| Fast | Faster | Good quality, lower resolution | Previewing ideas, high-volume batches |
Formats that do not use AI images (Slide C, Carousel C, Slide D, Carousel D) always render at full quality instantly since no image generation is involved.
Quality Scoring
Every piece of content generated by ViralToby is evaluated by the Quality Scoring Function (QSF) before it enters the pipeline. The QSF judges viral fitness and pattern compliance across five dimensions:
| Dimension | Weight | What It Measures |
|---|---|---|
| Structural Compliance | 25% | Does the content follow the expected format? Correct number of points, word limits, proper structure. |
| Pattern Familiarity | 20% | Does it match proven content patterns from your niche? Recognizable frameworks that audiences engage with. |
| Novelty | 20% | Is it fresh and not repetitive? Avoids rehashing recently published content. |
| Emotional Hook Strength | 20% | Does the hook grab attention? Uses proven psychological triggers (curiosity, surprise, urgency). |
| Plausibility | 15% | Is the content believable and accurate? Avoids outlandish claims that damage credibility. |
Score Thresholds
| Score Range | Action |
|---|---|
| 80-100 | Publish -- content meets quality standards and enters the pipeline |
| 65-79 | Regenerate -- content is tweaked and re-scored |
| Below 65 | Reject -- content is discarded and regenerated from scratch |
Quality scoring happens automatically. You do not need to configure it. The QSF ensures only high-quality content reaches your review pipeline.
Design Customization
Brand Colors and Fonts
Each brand has its own visual identity configured in its settings:
- Primary and secondary colors -- used for text, backgrounds, and accents
- Font selections -- brand-specific typography for thumbnails and text overlays
- Layout preferences -- how text is positioned on slides
Thumbnail Types
ViralToby offers two thumbnail designs, independent from the content format:
| Thumbnail | Style |
|---|---|
| Thumbnail A | Clean, modern layout with bold title text and brand accent colors |
| Thumbnail B | Alternative layout with different visual emphasis and composition |
Thumbnail type is set per brand in your brand configuration. You can change it at any time -- it only affects the visual rendering, not the content itself.
Content format (A or B) and thumbnail type (A or B) are independent axes. You can use any combination: Format A with Thumbnail B, Format B with Thumbnail A, and so on. This gives you four distinct visual styles without changing your content strategy.
Caption Generation
Every piece of content gets a platform-optimized caption generated by AI:
- Instagram: Longer captions with hashtags, emojis, and line breaks for readability.
- Facebook: Slightly different formatting optimized for Facebook's text display.
- YouTube: Title and description formatted for YouTube Shorts discovery.
- Threads: Concise text optimized for the conversational Threads format.
- TikTok: Short, punchy captions with trending hashtags.
- Bluesky: AT Protocol-compatible text posts.
Captions are generated from your Content DNA settings (CTA options, hashtags, brand personality) and tailored to each platform's best practices.
Music Selection
Music is a key component of Reels and video content. ViralToby supports three music modes:
| Mode | Description |
|---|---|
| Auto (weighted random) | ViralToby selects from its built-in music library using weighted random selection. Tracks that pair well with your niche are preferred. |
| Trending | Select from currently trending audio tracks. |
| User-uploaded | Upload your own music tracks to use in content. |
| None | No music -- silent video. |
When Toby generates content autonomously, it automatically selects appropriate music. For manual creation, you can choose the music source and specific track.
Ensure you have the rights to use any music in your content, especially user-uploaded tracks. ViralToby's built-in library consists of royalty-free tracks suitable for commercial use.
Batch Generation
When you click Generate All or Toby fills multiple buffer slots at once, ViralToby generates content in parallel across brands and formats using a thread pool. This means a 10-brand batch does not take 10x longer than a single-brand job — brands are processed concurrently.
Batch Size
Super admins can trigger batches of up to 10 jobs at once from the creation wizard. Standard users generate one job at a time per click.
Webhook-Accelerated Images
For AI image generation (Content B reels, Carousel B, Slide B), ViralToby uses a webhook callback rather than polling. When the image provider completes an image, it notifies ViralToby instantly via a signed webhook. This means:
- Image jobs complete as fast as the provider generates them — no fixed polling delay
- Multiple images in the same batch finish in parallel, not sequentially
- The pipeline updates in real time via Supabase Realtime as each image completes
Cross-Batch Deduplication
Even when multiple batches run in parallel, ViralToby prevents two concurrent jobs from generating the same title. This is enforced via reserve_title() — a server-side reservation that locks a title for a short window while the job runs. If two concurrent jobs produce the same title, one is regenerated with a new angle before entering the pipeline.
Content Deduplication
ViralToby has a built-in anti-repetition system that prevents duplicate or near-duplicate content from being published:
Title Deduplication
Before a new piece of content is finalized, ViralToby checks it against your recent content history. If the title is too similar to something recently published or scheduled, the content is flagged and regenerated.
Topic Diversity
The Content DNA topic system encourages variety. Even when Toby finds a winning topic category, the 35% explore ratio ensures it continues trying other topics to keep your feed diverse.
Quality Guard
Toby runs a Quality Guard sweep every tick cycle that inspects scheduled content and detects:
- Duplicate titles -- near-identical content that would bore your audience
- Caption duplicates -- repetitive captions across scheduled posts
- Slot collisions -- multiple posts scheduled at the same time
- Fallback content -- lower-quality content that was generated as a backup
Detected duplicates are automatically cancelled, and their time slots are freed up for fresh content.