An enterprise-grade, AI-native platform engineered to transform raw voice or text inputs into fully orchestrated, multi-scene cinematic advertisements at scale. Powered by agentic prompt intelligence, event-driven automation, and distributed cloud rendering, the system executes an end-to-end creative pipeline—from ideation and scene generation to compositing and final delivery—with precision and speed.
This platform converts a user’s Telegram voice message or text ad idea into a fully produced, cinematic 6-scene video advertisement using generative AI. It automatically:
- Accepts voice or text input from Telegram
- Transcribes speech using OpenAI
- Designs a six-scene ad narrative using GPT
- Generates each scene via Google Veo3
- Stores clips in Google Drive
- Indexes them in Airtable
- Stitches them into a full video using JSON2Video
- Delivers the final ad back to the user on Telegram
The system runs as a fully autonomous, agentic, AI-driven creative production pipeline. :contentReference[oaicite:1]{index=1}
- 🏷️ Project Title
- 🧾 Executive Summary
- 🧩 Project Overview
- 🎯 Objectives & Goals
- ✅ Acceptance Criteria
- 💻 Prerequisites
- ⚙️ Installation & Setup
- 🔗 API Documentation
- 🖥️ UI / Frontend
- 🔢 Status Codes
- 🚀 Features
- 🧱 Tech Stack & Architecture
- 🛠️ Workflow & Implementation
- 🧪 Testing & Validation
- 🔍 Validation Summary
- 🧰 Verification Tools
- 🧯 Troubleshooting
- 🔒 Security & Secrets
- ☁️ Deployment
- ⚡ Quick-Start
- 🧾 Usage Notes
- 🧠 Performance
- 🌟 Enhancements
- 🧩 Maintenance
- 🏆 Achievements
- 🧮 High-Level Architecture
- 🗂️ Project Structure
- 🧭 Live Demonstration
- 💡 Summary, Closure & Compliance
AI-Veo3-Ad-Generation-Platform is a production-grade AI Creative Automation Engine. It removes human intervention from the entire ad creation lifecycle by combining:
- Conversational AI
- Multimodal prompt engineering
- AI video synthesis (Veo3)
- Automated media storage
- Database indexing
- Final cinematic rendering
The system is triggered by Telegram and orchestrated entirely by n8n. :contentReference[oaicite:2]{index=2}
- Convert voice → ad → video with zero manual steps
- Enforce consistent cinematic structure across all ads
- Enable scalable AI-driven creative production
- Provide cloud-stored video assets and traceability
- Deliver output back to users instantly
| Area | Acceptance Requirement |
|---|---|
| Input Handling | System must accept both Telegram voice and text messages and route them correctly |
| Transcription | Voice messages must be transcribed into text with >95% accuracy |
| AI Script | GPT must generate a valid 6-scene JSON payload for Veo3 |
| Video Generation | Each scene must be rendered by Veo3 and stored successfully |
| Database | Airtable must persist all scene URLs with the campaign name |
| Final Video | JSON2Video must produce a Full-HD merged ad |
| Delivery | Final video must be sent to the originating Telegram user |
- n8n (self-hosted or cloud)
- Telegram Bot Token
- OpenAI API Key
- Veo3 (fal.ai) API Key & Secret
- Google Drive OAuth credentials
- Airtable Base & Table
- JSON2Video API Key
- Public HTTPS webhook URL
- Deploy n8n using Docker or Cloud
- Import veo3-ad-generator.json workflow
- Configure credentials:
- Telegram
- OpenAI
- Google Drive
- Airtable
- Veo3 (fal.ai)
- JSON2Video
- Create Airtable schema with fields: Name, voice_over, scene_link1 → scene_link6
- Create Google Drive folder for videos
- Set Telegram webhook to n8n trigger URL
| Service | Endpoint | Purpose |
|---|---|---|
| Telegram | getFile / sendVideo | Receive voice & deliver video |
| OpenAI | /audio/transcriptions | Speech to text |
| OpenAI GPT | /chat/completions | Ad scene generation |
| Veo3 | queue.fal.run/fal-ai/veo3 | Scene video generation |
| JSON2Video | /v2/movies | Final video stitching |
The platform uses Telegram as the UI layer.
| Layer | Description |
|---|---|
| Telegram Chat | User inputs voice or text |
| Telegram Video Player | Final ad playback |
No web frontend is required. All state, scenes, and progress is handled by n8n + Airtable.
| Code | Meaning |
|---|---|
| 200 | Successful video generation |
| 400 | Invalid input or missing prompt |
| 401 | Invalid API keys |
| 429 | Rate limit reached |
| 500 | AI service or Veo3 failure |
| Capability | Description |
|---|---|
| Voice Intake | Telegram accepts audio messages and extracts voice files |
| Speech-to-Text | OpenAI Whisper transcribes spoken ad ideas |
| Ad Intelligence | GPT-based AdScript AI creates 6-scene narrative JSON |
| Video Generation | Each scene rendered using Veo3 via fal.ai queue API |
| Cloud Storage | All scenes stored to Google Drive |
| Data Index | Scene links persisted into Airtable |
| Final Render | JSON2Video stitches 6 scenes into a full HD movie |
| Delivery | Telegram sends final video + URL back to the user |
Telegram → n8n → OpenAI → Veo3 → Google Drive → Airtable → JSON2Video → Telegram
[User] | [Telegram] | [n8n Orchestrator] | [OpenAI (Transcribe + Script)] | [Veo3 Generator] | [Google Drive Storage] | [Airtable Scene Index] | [JSON2Video Movie Engine] | [Telegram Delivery]
- User sends voice/text to Telegram
- n8n detects message type
- Voice → OpenAI Whisper → text
- Text → GPT → 6-scene JSON
- Each scene sent to Veo3
- Veo3 returns video URLs
- Videos saved to Google Drive
- Links stored in Airtable
- JSON2Video builds final movie
- Telegram sends the final ad
| ID | Area | Command | Expected Output | Explanation |
|---|---|---|---|---|
| T01 | Telegram | Send voice message | n8n triggered | Validates webhook |
| T02 | OpenAI | Submit audio | Transcript returned | Speech pipeline |
| T03 | Veo3 | Submit prompt | Video URL | Scene generation |
| T04 | JSON2Video | Create movie | MP4 URL | Final render |
- All six scenes are verified in Airtable
- Google Drive contains all video clips
- Final movie plays successfully
- Telegram receives the output
- Postman – API checks
- Telegram Bot console
- n8n Execution Logs
- Airtable record viewer
- Google Drive media preview
| Issue | Cause | Fix |
|---|---|---|
| No video | Veo3 timeout | Increase wait nodes |
| Missing scenes | Airtable mapping | Fix upsert schema |
| Telegram failure | Bot token expired | Regenerate token |
- All API keys stored in n8n credentials vault
- Telegram Webhooks use HTTPS
- No secrets are hardcoded
Recommended deployment: n8n on VPS or Cloud with HTTPS + public IP.
Flow:Telegram → HTTPS → n8n → Cloud APIs → User
- Send voice message to Telegram bot
- Wait ~2–5 minutes
- Receive AI-generated ad video
- Best results with 5–20 sec voice prompts
- Describe product, audience, and tone
- Parallel Veo3 scene generation
- Async waits avoid blocking
- Cloud-based scaling via fal.ai
- Multi-language ads
- Brand voice presets
- Auto subtitles
- Bulk campaign generation
- Add web dashboard
- Add billing system
- Brand asset library
- Fully autonomous ad creation
- AI cinematic storytelling
- Cloud-native video factory
AI-VEO3-AD-GENERATION-PLATFORM │ ├── assets │ └── diagrams │ ├── system-architecture-1.png │ ├── system-architecture-2.png │ ├── system-architecture-3.png │ ├── system-architecture-4.png │ ├── system-architecture-5.png │ ├── system-architecture-6.png │ ├── system-architecture-7.png │ └── system-architecture-8.png │ ├── docs │ ├── architecture.md │ ├── data-flow.md │ ├── execution-phases.md │ ├── services.md │ └── setup.md │ ├── workflows │ └── veo3-ad-generator.json │ ├── .env.example ├── .gitignore └── README.md
Voice/Text ↓ Telegram Bot ↓ n8n Intake Router ↓ AI Transcription ↓ AI Creative Director ↓ 6× Veo3 Video Jobs ↓ Cloud Storage ↓ Scene Database ↓ Final Video Composer ↓ User Delivery
- Open Telegram
- Send voice message to bot
- Observe n8n workflow execution
- Watch scenes appear in Google Drive
- Final video arrives in Telegram
This platform represents a **next-generation AI creative production system** capable of generating full cinematic advertisements without any human intervention. It complies with:
- Cloud-native orchestration principles
- Zero-trust API security
- Event-driven automation design
- AI-first creative workflows
This architecture is scalable, monetizable, enterprise-ready, and compliant with modern AI SaaS production standards. The system can be commercialized as an AI Ad Factory, Creative Automation API, or White-Label Marketing Engine. :contentReference[oaicite:3]{index=3}