AI YouTube Investing Influencer (3-Part)

A production-grade, 277-node automation system that transforms raw financial data into polished YouTube market analysis scripts. This three-part pipeline demonstrates what is possible when you combine multiple AI models, vector-store retrieval, and relentless orchestration logic within n8n. It is the kind of workflow that replaces an entire content production team.

Nodes	Category	Complexity	Key Integrations
277	Content Creation & AI	Expert	Finviz, Alpha Vantage, OpenAI, Anthropic, Gemini, Mistral, Brave Search, Qdrant, Google Sheets

Prerequisites

• Finviz account (Elite plan recommended for full screener API access)
• Alpha Vantage API key (free tier available but rate-limited)
• OpenAI API key for scoring and classification agents
• Anthropic API key for research synthesis agents
• Google Gemini API key (used in select pipeline stages)
• Mistral API key (used in select pipeline stages)
• Brave Search API key for web research
• Qdrant instance (cloud or self-hosted) for vector store RAG
• Cohere API key for embeddings
• Redis instance for multi-agent memory
• Google Sheets OAuth2 credentials for data handoff and script storage
• Knowledge level: Expert -- requires strong understanding of multi-workflow orchestration, financial APIs, rate limiting, and multi-model AI architecture

Estimated Setup Time

2-3 hours including financial API setup, vector store initialization, Redis configuration, and three-part workflow wiring.

What It Does

Part 1 -- Stock Scanner. The pipeline begins by scanning over 1,000 publicly traded stocks through financial data APIs including Finviz and Alpha Vantage. Each stock passes through a series of Code nodes that calculate technical indicators, filter by configurable criteria (market cap, volume, sector), and normalize the data into a consistent format. AI agents then evaluate the filtered results, scoring each stock on momentum, fundamentals, and sentiment. The top candidates are written to a master Google Sheet that serves as the bridge to Part 2.

Part 2 -- Research Engine. For each high-scoring stock from Part 1, the Research Engine fires off parallel web searches using Brave Search and retrieves relevant context from a Qdrant vector store populated with historical analyses. Cohere embeddings power the RAG pipeline, ensuring the AI retrieves the most semantically relevant prior research. All findings are synthesized by an Anthropic Claude agent into structured research briefs, complete with citations and confidence assessments.

Part 3 -- Script Generator. The final stage consumes the research briefs and transforms them into broadcast-ready YouTube scripts. Multiple AI agents handle different sections: the hook, the analysis body, counterarguments, and the call to action. Redis memory maintains context across the multi-agent conversation, ensuring each section builds logically on the last. The finished script lands in a dedicated Google Sheet, tagged with metadata and ready for recording.

Workflow Diagram

Node Breakdown

Triggers

• Schedule Trigger
• Manual Trigger

AI & Language Models

• AI Agent (7)
• OpenAI
• Anthropic
• Gemini
• Mistral
• Brave Search
• Qdrant Vector Store
• Cohere Embeddings
• Redis Memory

Data & Integrations

• HTTP Request (20+)
• Code (15+)
• Google Sheets (20+)
• Merge
• IF / Switch

Step-by-Step Flow

1. Schedule Trigger fires Part 1 on a configurable cadence (daily or weekly).
2. HTTP Request nodes pull stock data from Finviz screener and Alpha Vantage APIs for 1,000+ tickers.
3. Code nodes calculate technical indicators (RSI, MACD, moving averages) and filter stocks by configurable thresholds.
4. AI Agents score each filtered stock on momentum, fundamentals, and market sentiment using OpenAI models.
5. Google Sheets receives the ranked stock list, serving as the handoff point to Part 2.
6. Part 2 triggers on new rows. For each stock, Brave Search executes web searches in parallel.
7. Qdrant retrieval pulls semantically similar prior analyses from the vector store using Cohere embeddings.
8. Anthropic Claude agent synthesizes all research inputs into a structured brief with citations.
9. Part 3 triggers on completed research briefs. Multiple AI agents draft different script sections (hook, analysis, counterarguments, CTA).
10. Redis Memory maintains conversational context across agents so each section builds on the previous one.
11. Merge nodes combine all script sections into the final output.
12. Google Sheets stores the finished script with metadata tags for easy retrieval and recording scheduling.

What Sets It Apart

This workflow stands apart on four dimensions:

• Scale. Processing 1,000+ stocks per execution is not trivial in n8n. The batching logic, rate limiting, and error handling required to reliably hit financial APIs at this volume represents serious orchestration engineering.
• Multi-model AI. Rather than relying on a single LLM, the pipeline uses OpenAI for scoring and quick classification tasks while deploying Anthropic Claude for long-form synthesis and nuanced research summarization. Each model is used where it excels.
• RAG-powered research. The Qdrant vector store accumulates institutional knowledge over time. Every execution makes future executions smarter because prior analyses become retrievable context.
• Fully automated content pipeline. From raw market data to a finished YouTube script, no human intervention is required. The workflow handles data acquisition, filtering, research, writing, and organization end to end.

Cost Considerations

• Finviz Elite: ~$25-40/month for full screener access; free tier is heavily limited
• Alpha Vantage: Free tier allows 25 requests/day; premium starts at $50/month for higher volume
• OpenAI: Moderate cost per execution due to scoring 50-100+ stocks; estimate $1-5 per full pipeline run
• Anthropic Claude: Research synthesis is token-heavy; estimate $2-8 per full run depending on stock count
• Brave Search: Free tier (1 req/sec, 2,000/month); paid plans from $5/month
• Qdrant Cloud: Free tier (1GB); paid tiers for larger knowledge bases
• Cohere: Free trial tier available; production usage billed per embedding
• Redis Cloud: Free tier (30MB); sufficient for session memory
• Total estimated cost per full 3-part run: $5-20 depending on stock volume and model usage

Ideal For

• Financial content creators producing regular market analysis videos
• Investment newsletter writers who need data-driven research at scale
• Fintech startups building automated content around market data
• Media companies covering financial markets across multiple channels

Tip: Key Techniques Used in This Workflow

• Multi-workflow orchestration -- Three separate workflows communicate through Google Sheets as a shared data layer, enabling independent scaling and debugging of each stage.
• Parallel API batching -- Code nodes manage rate limits and batch HTTP requests to stay within API quotas while maximizing throughput.
• Multi-model routing -- Different AI models handle different task types based on their strengths, a pattern that improves both quality and cost efficiency.
• Accumulative RAG -- The vector store grows with each execution, creating a compounding knowledge advantage over time.
• Redis-backed multi-agent context -- Redis Memory nodes allow multiple AI agents to share conversational context within a single execution, enabling coherent multi-section document generation.

Download

The workflow.json file in this folder can be imported directly into your n8n instance via Settings > Import Workflow.