TrendScout is a lightweight AI-powered assistant that analyzes YouTube video titles to uncover packaging patterns, emotional hooks, and formatting styles used in top-performing content.
This tool was designed to test whether simple, repeatable workflows could help creators generate title insights without manually reviewing dozens of videos. It forms the first proof-of-concept (POC) for a larger visual analysis tool, ThumbScout, which will later include GPT-4 Vision.
⸻
- 🔍 Scrapes titles using SERP API (Google → YouTube engine)
- 🧠 Analyzes titles for emotional appeal, formatting, and clarity
- 📬 Compiles results into a CSV + emails the report to the user
- ✅ Fully automated workflow via n8n, including error handling
⸻
- Project Name: TrendScout
- Creator: Ros Talbot
- Date: July 31, 2025
- Version: MVP v0.1
- Build Time: ~2 hours
TrendScout successfully ran a complete agent-style pipeline:
- Live search → LLM analysis → CSV conversion → automated email delivery
It’s designed for:
- YouTube creators
- Content strategists
- Thumbnail/title testers
🟢 Low-lift, high-utility, and ready for iterative upgrades.
View full evaluation notes ▶ (linked if broken out later)
⸻
- Automation: n8n
- LLM: GPT-3.5 and GPT-4 (via OpenAI node)
- Data Source: SERP API (YouTube via Google)
- Delivery: Native email node (SMTP)
- Output: CSV + Markdown
⸻
- Add structured parsing (Markdown, JSON)
- Include original video links, publish dates, and views
- Batch analyze multiple queries at once
- Add scheduled triggers (e.g. weekly trend pull)
- Expand into GPT Vision for full ThumbScout build
⸻
-
ThumbScout → [FUTURE PROJECT LINK]
TrendScout served as the prototype for title analysis inside this larger vision-based agent.
-
Vibe Agent Series → [Parent Page Coming Soon]
TrendScout is part of a micro-agent portfolio exploring creative, lightweight AI tools built without traditional dev teams.
⸻
DM @RosTalbot or fork this agent and make it your own!