A next-generation, enterprise-grade AI system that transforms fragmented competitor ad data into high-performance, brand-safe video creatives through a fully orchestrated, agentic pipeline.
AI Ad Creative Strategist is an enterprise-grade, agentic AI system that automates the full lifecycle of high-performance advertising creative generation. The platform continuously monitors competitor ads, identifies winning creatives, extracts creative intelligence, adapts strategies to a target brand, and generates new AI-powered video advertisements using advanced multimodal models.
The solution replaces manual ad research, creative strategy formulation, and initial video production with a deterministic, auditable, and scalable workflow.
- 🧩 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 Testing Tools
- 🧯 Troubleshooting & Debugging
- 🔒 Security & Secrets
- ☁️ Deployment
- ⚡ Quick-Start Cheat Sheet
- 🧾 Usage Notes
- 🧠 Performance & Optimization
- 🌟 Enhancements & Features
- 🧩 Maintenance & Future Work
- 🏆 Key Achievements
- 🧮 High-Level Architecture
- 🗂️ Project Structure
- 🧭 How to Demonstrate Live
- 💡 Summary, Closure & Compliance
This project implements a five-phase AI advertising pipeline orchestrated through a workflow automation engine. It integrates external ad intelligence APIs, multiple AI models, persistent data storage, and human-in-the-loop controls.
The system is designed to be:
- Fully automated yet review-gated
- Data-driven and repeatable
- Scalable across brands and markets
- Auditable and compliant
- Automate competitor ad discovery
- Identify statistically winning creatives
- Extract reusable creative intelligence
- Adapt strategies to a new brand identity
- Generate production-ready AI video ads
- Reduce time-to-creative from weeks to minutes
- Workflow runs end-to-end without manual intervention
- Only qualified ads are processed
- Generated creatives are brand-aligned
- All outputs are logged and traceable
- No secrets are committed to source control
- Node-based workflow orchestration platform
- API access to ad intelligence provider
- Access to multimodal AI models
- Persistent datastore (Airtable)
- Team communication platform (Slack)
- Clone the repository
- Configure environment variables
- Import workflow JSON
- Bind credentials
- Populate brand data
- Enable scheduled trigger
The AI Ad Creative Strategist integrates with multiple external APIs to enable end-to-end creative intelligence, adaptation, and generation. Each API plays a distinct role in the pipeline and is invoked at a specific phase of execution.
| API / Service | Phase Used | Purpose | Data In | Data Out |
|---|---|---|---|---|
| Foreplay Public API | Phase 1 | Competitor ad discovery | Brand ID, filters | Ad metadata, video URLs |
| Gemini Multimodal API | Phase 3 | Creative intelligence extraction | Video URL, transcript | Scene analysis, hooks, CTA logic |
| Anthropic Claude API | Phase 4 | Brand strategy adaptation | Creative blueprint, brand data | SORA-optimized prompt |
| SORA 2 Video API | Phase 5 | Video ad generation | Structured prompt | Generated video asset |
| Slack Webhook API | Phase 5 | Team notification | Status payload | Message delivery |
- All API calls are synchronous within their execution phase
- Rate limiting is controlled via loop iteration boundaries
- Failures are isolated per ad item
- Retries are delegated to workflow-level controls
This project is backend-first. The UI consists of:
- Workflow visualization canvas
- Airtable dashboards
- Slack notifications
Styling changes are managed at the brand data level rather than UI code.
| Code | Meaning |
|---|---|
| 200 | Successful execution |
| 400 | Invalid input or configuration |
| 401 | Authentication failure |
| 500 | External service failure |
The platform delivers a comprehensive feature set designed for modern performance marketing, creative operations, and AI-driven growth teams.
- Automated competitor ad ingestion with zero manual research
- Runtime-based qualification to identify proven winning creatives
- Second-by-second video and narrative intelligence extraction
- Brand-safe creative strategy adaptation using AI agents
- Automated cinematic video generation using structured prompts
- Full audit trail of inputs, transformations, and outputs
- Human-in-the-loop approval workflow
- Failure isolation and graceful degradation
- Cost control via deterministic filtering
- Extensible phase-based architecture
- Reduced creative ideation time from weeks to minutes
- Consistent creative quality across campaigns
- Scalable multi-brand execution
- Repeatable creative experimentation
| Layer | Technology | Role |
|---|---|---|
| Orchestration | n8n | Workflow automation and control |
| Ad Intelligence | Foreplay API | Competitor ad discovery |
| Analysis AI | Gemini | Multimodal creative analysis |
| Strategy AI | Claude | Brand adaptation and reasoning |
| Generation AI | SORA 2 | Video ad generation |
| Persistence | Airtable | Audit logs and brand data |
| Notification | Slack | Human review signaling |
[ Daily Scheduler ]
|
v
[ Foreplay API ]
|
v
[ Qualification Filter ]
|
v
[ Loop Controller ]
|
v
[ Gemini Analysis ]
|
v
[ Claude Adaptation ]
|
v
[ SORA 2 Generator ]
|
v
[ Airtable Logs ] ---> [ Slack Notifications ]
- Scheduled trigger initiates a new execution context
- Competitor ads are fetched from the ad intelligence API
- Ad metadata is normalized and prepared for evaluation
- Ads are filtered using runtime qualification rules
- Each qualified ad enters a controlled processing loop
- Video transcripts are aggregated and cleaned
- Multimodal AI analyzes creative structure and messaging
- Creative intelligence is persisted for traceability
- Brand context is loaded from internal datastore
- AI agent adapts winning strategy to brand constraints
- Structured video generation prompt is produced
- Video creative is generated using SORA 2
- Outputs are logged and notifications are dispatched
- Errors are scoped to individual ads
- Workflow continues processing remaining items
- Execution logs capture error context
| ID | Area | Command | Expected Output | Explanation |
|---|---|---|---|---|
| T-01 | Ingestion | Manual run | Ads fetched | Validates API access |
| T-02 | Filtering | Manual run | Qualified ads only | Ensures logic correctness |
All workflow phases were validated through controlled manual executions and persisted data inspection.
- Workflow execution logs
- Airtable record inspection
- Slack message verification
- Inspect node-level execution logs
- Validate intermediate data objects
- Re-run workflow with a single ad
- All secrets stored externally
- .env.example provided
- No credentials in repository
The project is designed for backend execution, while documentation and dashboards can be deployed using modern frontend hosting platforms.
- Workflow Engine: Self-hosted or managed
- Documentation: Vercel (static deployment)
- Stateless execution model
- No runtime frontend dependency
- Separation of compute and presentation
- Import workflow
- Bind credentials
- Enable trigger
- Review Slack output
- Generated ads are paused by default
- Human review is mandatory
- Batch processing reduces cost
- Filtering minimizes AI usage
- Stateless execution ensures scalability
- Multi-brand parallel execution
- Creative A/B feedback loop
- Auto-scoring creatives
- Model upgrades
- New ad platforms
- Creative performance feedback
- End-to-end AI creative automation
- Brand-safe generative ads
- Production-ready architecture
[ Time Trigger ]
|
v
[ Competitor Ad Source ]
|
v
[ Qualification Layer ]
|
v
[ Intelligence Extraction ]
|
v
[ Brand Strategy Layer ]
|
v
[ Creative Generation ]
|
v
[ Audit & Notification ]
AI-AD-CREATIVE-STRATEGIST/ ├── assets/ │ └── diagrams/ │ ├── 01-workflow-trigger-and-ad-ingestion.png │ ├── 02-winning-ad-filtering-and-analysis.png │ ├── 03-brand-adaptation-and-prompt-generation.png │ └── 04-video-generation-and-delivery.png ├── docs/ │ ├── architecture-overview.md │ ├── workflow-phases.md │ ├── data-flow.md │ └── setup-guide.md ├── workflows/ │ └── ai-ad-creative-strategist.json ├── .env.example ├── .gitignore └── README.md
- Open workflow canvas
- Run manual execution
- Show Airtable records
- Show Slack notification
The AI Ad Creative Strategist represents a compliant, auditable, and production-ready implementation of agentic AI for marketing operations.
- Separation of secrets from code
- Human approval gates
- Deterministic execution paths
- Complete audit trails
This system demonstrates how modern AI can be operationalized responsibly to augment creative teams, reduce costs, and improve advertising effectiveness without compromising governance or control.