Our Research
Work from the Kepler team, published as we do it. Financial research built on our own platform, where every number is pulled by code from the source filing so you can verify anything we claim. And engineering research on how we build Kepler itself.
Financial Research
Engineering Research
Better language models won't fix hallucination. They'll make it quieter.
A number's digits don't contain its fact. Better models make wrong numbers rarer and harder to see; the fix is a provenance layer where the model names facts by reference and never retypes the digits.
Stop Paying to Forget
Most document AI pays to reread documents for every question. We read each one once, keep what we read, and answer the next question with ordinary software.
Reclassified to Conform
194 of 200 companies reshaped a financial statement in ways XBRL tags can't follow. Our new paper shows how to recover what changed from the filings' own arithmetic.
From the Blog
All postsRun this on your own coverage
We build these notes and models with Kepler, the same platform investment teams use on their own data. Ask a question, get an answer with every number traced to its source.