Property Analysis Tool
Core Idea
A website or app that helps users analyse Singapore property prices by extracting insights from YouTube property experts using AI, combined with real transaction data.
State of Thinking
The idea has two distinct parts that need to work together:
- Knowledge extraction — use NotebookLLM to process and distil insights from YouTube property expert content, turning unstructured video commentary into structured analysis
- Transaction data — layer real property transaction data on top of that extracted knowledge to ground the analysis in actual market reality
The combination is the interesting part. YouTube property experts have opinions and frameworks; transaction data has facts. Neither is sufficient alone.
Still very early — the knowledge extraction side feels more solved (NotebookLLM is a known tool) but the transaction data source is an open problem.
Thinking Trace
- Property content on YouTube is abundant but unstructured and hard to normalise across different creators
- NotebookLLM feels like a natural fit for extracting and synthesising that content
- The missing piece is where to get recent transaction data — URA has public data but not sure if it’s accessible in a useful format
- Not sure yet if this is a tool for personal use or something to build for others
- The AI angle makes this feel more tractable than building a pure data product from scratch
What changed?
— skip, v1 —
Why it changed?
— skip, v1 —
Evidence / Signals
- YouTube property expert content is abundant in Singapore — large amount of raw material to work with
- URA releases transaction data publicly — may be a viable data source worth investigating
Open Questions
- Where to get recent Singapore property transaction data — URA, SRX, other sources?
- How reliable and consistent is NotebookLLM for structured knowledge extraction at scale?
- Who is the target user — personal tool or public product?
- How do you handle conflicting opinions between different YouTube experts?
- What does the actual user experience look like — search by area, compare properties, track trends?
Connections
- (none yet — first node in this tree)
Changelog
- v1 — Initial node