Skip to main content

​
How To Ask Questions

Early Bird works best when users ask for analysis, comparison, ranking, clustering, or timelines.

​
Good question patterns

  • Compare two or more groups
  • Rank organizations or programs by a metric
  • Summarize trends over time
  • Filter by tag family, category, region, or year
  • Ask for a concise explanation of what the data suggests

​
Good examples

  • Compare top university-backed programs in Hong Kong by project volume and category mix.
  • Find AI or climate-related projects from accelerator-backed organizations in 2024.
  • Which organizations have the most active or upcoming application windows right now?
  • Summarize the most common market, domain, and technology tags across recent projects.
  • Identify programs with high project density and explain what kinds of startups they surface.

​
Questions to avoid by default

  • Give me the entire database.
  • Fetch every page for every entity.
  • Return every row with every field unless explicitly necessary.
These requests are possible in principle, but they are intentionally discouraged because Early Bird is designed to be conservative with infrastructure usage.
  1. Read the docs or data catalog first.
  2. Start with a summary endpoint.
  3. Fetch one small paginated result set only if needed.
  4. Stop early if the question can already be answered.
  5. If the data is partial or truncated, say so clearly.