kayetanaTriage claims faster. Keep data in-house.
Claims, policies, policy wordings and expert reports: every day, a lot of text lands on an insurer's desk. kayetana helps you sort and look things up, on a server you run yourself.
Pre-sort incoming claims automatically.
Define labels like auto, home contents, liability or follow-up, and add a few examples per label. kayetana assigns new claims and shows a confidence score per label. That way, every claim reaches the right team sooner.
- Examples right in the request
- One or more labels per claim
- Your own classifier trained on your data
Where kayetana helps in day-to-day insurance work.
Query policy wordings
Ask whether a loss is covered and see the relevant clause as the source.
Compare versions
Put old and new terms side by side and get a summary of what changed.
Summarize expert reports
Load long adjuster reports as PDFs and pull out the key points.
Answer customer questions
Draft replies to policy questions, with team templates for a consistent tone.
Capture phone calls
Record first notice of loss calls and file them as text with timestamps.
Extract structured data
Use JSON mode to turn free text into fields like date, location and cause of loss.
Every answer comes with the clause behind it.
Load policy wordings, product descriptions and internal procedures as PDF or Word files. kayetana answers only from these texts and adds a source number after every sentence. You check the passage yourself before you decide.
- PDF, Word, RTF, ODT and HTML
- A source list under every answer
- Also via API with citations
Four steps to a pilot.
Pick a line of business
For example home contents or auto, with a manageable volume of claims.
Define labels and examples
Agree with the business team on how claims should be sorted.
Load your policy wordings
Provide the current versions as documents for answers with citations.
Review the results
Evaluate assignments and answers together with your claims handlers.
See kayetana with examples from your line of business.
In the demo, we sort sample claims and query a policy wording with citations.
Frequently asked questions
Does kayetana store claims or customer data?
No. Prompts and responses are never stored. Only time, endpoint, model, status, tokens and duration are logged.
Does kayetana decide on claims settlement?
No. kayetana sorts, summarizes and shows sources. Your team always makes the decision.
Can I train a classifier on my own claims?
Yes. From a labeled dataset, you build your own classifier "clf-name" on top of an embedding model.
Does kayetana meet insurance regulations?
We can't promise that. kayetana keeps data on your server. Review the requirements with your compliance team and data protection officer.