Ballpark

About

You're reading that a cybersecurity startup raised a $28M Series B, a biotech enrolled 180 patients in its Phase II trial, or a hardware brand runs 34% gross margins. Is that big? Small? Normal for that field?

If you don't live in the industry, you have no reference frame — and your existing intuition often misleads you. A 34% margin is alarming for SaaS but healthy for physical goods. A $28M Series B is small for AI but generous for security tooling.

Ballpark gives you a quick directional sense, in place, without leaving the page.
Click any underlined number and a small card appears with a one-line verdict (is this big, small, or typical?), the reference class it's being compared against, and one or two grounded comparisons to anchor your intuition. A green "Searched" indicator means the answer was checked against live web search; a gray "From memory" indicator means it came from the model's training data — useful for direction, not for exact figures.

Ballpark is for frontier readers keeping up with new tech and markets, domain crossers moving into unfamiliar industries, and analysts who need a fast first read before they dig in.
How it works:

Free to use, fully open source (MIT license)
Requires your own Anthropic API key — calls go directly from your browser to Anthropic, no middleman, no Ballpark server
Reads page content only when you click a number, never in the background
Repeats are cached locally and cost nothing

This is a calibration tool, not a fact-checker. Use it to orient quickly, not to verify specifics.

Privacy policy: https://stephenieliew.com/ballpark/privacy
Source code: https://github.com/stephenie-l/ballpark