Project / 016Case study
FanFuel: Turning Social Comments into Content Ideas
An AI intelligence engine that turns thousands of raw Reddit, Instagram and TikTok comments into 12+ structured content ideas in under 10 seconds.
- AI systems
- Product
- Role
- Founder
- Year
- 2026
- Language
- TypeScript
- Platform
- Web
- Reading
- 2 min
- Stack
- LLMs
- TikTok
The problem
Creators spend hours reading comment sections for ideas. FanFuel does that research: it reads the comments for them and returns structured content ideas.
The system
4 components · 3 connections · 1 request flow
Raw comments from three social platforms are normalised by a parsing layer, then an LLM turns them into structured content ideas.
- External service
- Worker
- Model / AI
- Service
Select a component to see its role, the technology behind it and what it connects to, or follow a request through the system hop by hop.
Component reference4 components · 1 flow · as text
Components · in request order
- External service · layer 01Social platforms
- Thousands of raw comments
- Reddit, Instagram and TikTok, each with a different data schema.
- TechReddit, Instagram, TikTok
- ToParsing layer (raw comments)
- Worker · layer 02Parsing layer
- Normalises every schema
- Keeps data consistent for downstream LLM processing.
- TechSchema mapping
- FromSocial platforms (raw comments)
- ToLLM processing (normalised comments)
- Model / AI · layer 03LLM processing
- Turns comments into ideas
- TechLLM
- FromParsing layer (normalised comments)
- ToContent ideas (structured ideas)
- Service · layer 04Content ideas
- 12+ structured ideas in under 10 seconds
- TechStructured output
- FromLLM processing (structured ideas)
Request flows
Comments to ideas
- Step 1: Social platforms
- Step 2: Parsing layer
- Step 3: LLM processing
- Step 4: Content ideas
Decisions & trade-offs
2 decisions
- D01
Normalise before the model sees anything
A parsing layer handles the different data schemas of Reddit, Instagram and TikTok, so downstream LLM processing always gets consistent input.
- D02
Compress thousands of comments into a handful of ideas
The output is a short list of structured content ideas, not a summary to read through.
Outcomes
- 10x reduction in content research time for creators.
- 12+ structured content ideas from thousands of comments in under 10 seconds.
RoleFounder
The write-up
2 min read
FanFuel is an AI-driven intelligence engine for creators. It takes thousands of raw social media comments and turns them into 12+ structured content ideas in under 10 seconds, a 10x reduction in content research time.
How it works
A parsing layer handles the different data schemas of Reddit, Instagram and TikTok and keeps them consistent for downstream LLM processing. The model then turns the normalised comments into structured ideas.
Status
I founded and built FanFuel. More detail will be added here.