Reconstruct a company.
Hide a company’s figures. Learn to infer them from the surrounding economy. Reveal the answer and learn from the difference.
NEXT-GEN AI, SPECIALISED IN FINANCIAL MARKETS
Today’s AI was trained to predict the next word, and its intelligence is bounded by that training: it learned how the world is described, not how it works. Next-gen AI is trained to reconstruct the world, inferring each withheld piece from everything that surrounds it. Kenshin is next-gen AI, trained by that method on one subject: financial markets. Where today’s AI has read about markets, Kenshin has reconstructed them.
Kenshin is research, for people who want to understand companies and markets more deeply.
See how it’s trained

“Should I be worried?”
“Is this overvalued?”
“What am I missing?”
Explore the business behind a stock, the assumptions in its price, and what could change the picture.
A closer look at ambitious opportunities: the case for the upside, the risks, and what would need to go right.
Put a headline in context. Understand the connections between an event, a company and the wider economy.
These are features of our working prototype. Its current answers and research articles are clearly labeled samples.
Read a company through the world around it. Compare the market with a dated view of its value. Explore what changes when the assumptions change.
Try the example below. All figures are illustrative; no live company analysis is being performed.
Specialist AI agents investigate evidence, develop mathematical ideas and test explanations through computational experiments. Earlier discoveries return to the research, where they meet new information and ideas from other fields.
A connection found in one investigation can become the starting point for another. Findings are tested, combined and developed further—carrying understanding forward.
Hide a company’s figures. Learn to infer them from the surrounding economy. Reveal the answer and learn from the difference.
Bring historical records and synthetic economies into the same inquiry. Vary conditions, test explanations and examine where understanding carries over.
Evaluate emerging work in AI, mathematics and science. Connect it with earlier findings, adapt useful methods and test them against economic questions.
AI advances informing this perspective: Mathematical reasoning ↗Protein structure prediction ↗
Earlier findings become the starting point for new understanding.
Conceptual animation of our research approach. It does not display live server activity.
RESEARCH IN ACTION · REAL FINANCIAL RECORDS
Patterns learned across company histories give a few new observations a much larger context. Explore how four selected financial readings reconstruct a wider picture—and see how that estimate compares with what companies reported.
Explore the reconstructionInteractive historical research example · six quarters
Read a few signals. Reconstruct. Compare with reality.
Inspired by insect sensing
An insect can respond to a gust without forecasting its arrival. Kenshin’s research explores a related question: which economic signals matter, and what can they reveal about the wider system?
Learned relationships give a few observations a larger context.
The connection to KenshinConcentrating training and research on economics and finance creates possibilities beyond what a general-purpose model delivers on its own. Kenshin brings that sustained focus to the financial questions you want to explore.
COMING TO IPHONE
We’re preparing Kenshin for its first iPhone beta.
A download invitation will appear here when testing opens. Android is planned.
In development