Traditional screeners
Only filter candidates by the metrics they provide.
Your stock-picking,
handed straight to AI.
Katana narrows companies with its own metrics, and AI reads annual reports to judge qualitative conditions. For complex theses and company profiles, the selection you used to do one company at a time, AI now runs across the Japanese market.
01 / Compare
How Katana differs
Asking generative AI
It does not actually screen every listed company against your intent.
Existing AI screeners
Held back by computing cost, they could not run qualitative screens across the whole market.
In the end, even after narrowing, it was one company at a time.
You had to read company information and check whether each one fit your approach.
Nor could you search all Japanese stocks for a free-form condition like “companies quietly running a semiconductor business.”
With Katana, from narrowing to judging to analysis, end to end.
Katana
Stock selection that took hours and still missed companies,across about 3,800 listed companies,
done in minutes.
02 / Features
Katana's three features.
Custom metrics
1,000+ metrics, plus metrics you build with your own formulas. Turn an investment thesis straight into quantitative conditions.
e.g. how much ordinary income rises if rates go up 1%
AI judgment
AI reads annual reports and judges qualitative conditions, company by company. From ~3,800 companies and 70M+ characters of text, it selects the stocks that fit your thesis.
e.g. whether profits include one-off factors
MCP
Your AI can use Katana. Describe the companies you want, and the AI completes even complex conditions.
e.g. “Find high-dividend stocks that are recession-resistant, cheap, and on a progressive dividend policy”
Custom metrics, 10 years of financials, and MCP are free to use. You pay only for the AI judgment you use (on the web, this includes breaking your description into conditions). New accounts get a small free allowance.
03 / Example
From one sentence, across Japan.
“Find high-dividend stocks that are recession-resistant, cheap, and on a progressive dividend policy”
Sent as this one sentence (in Japanese) to Claude connected to Katana via MCP. Data as of October 28, 2026.
~3,800listed companies covered
133narrowed with Katana's metrics
2judged by AI reading annual reports
About two minutes, end to end.
See more MCP use casesKatana keeps sharpening.
Katana lets you search every listed company in Japan with your own hypothesis, not only by financial numbers but by the business described in annual securities reports. AI technology, advancing every day, made this possible. We will keep bringing new technology into Katana and keep sharpening it.
Why I Built KatanaThe background and thinking behind Katana Screener. Open the screener