Tesora: Frontier AI for Actuaries

Complete complex actuarial work in half the time.

Precise actuarial work, finally trusted to AI

Get actuarial work done fast with reproducible audit trails at a fraction of the cost.

Foundation Capital

Y Combinator

What we ship

We built the infrastructure around the models.

Each one takes over a job that used to eat somebody’s week, every month.

Between the model and your desk

AI that can’t hallucinate

There are hundreds of checks between the model and your desk. If a number cannot say which page it came from, you never see it.

Underneath the desk your team works at, every agent has checks to clear before its answer counts as finished. They exist because the one time a model is confidently wrong, it is a rate filing, a reserve opinion, or a number somebody testified to.

Built for enterprise security

Firewalled within your own infrastructure so no one sees your data

Your data stays in a private cloud of your own. Nothing you put in joins a shared index, and none of your work passes through anybody else’s context window.

FAQ

What people actually ask us.

From chief actuaries, from pricing teams, and from the IT and risk officers who review us alongside them.

Bring us the question you are dreading

A tangled web of formulas, a stale memo, a rater your team can't get off Excel. Make it hard for us.

Customers

Start with one team. Keep building.

Single-tenant

Your own private VPC, on hardware nobody else is running on, with no route out to the public internet.

Zero data retention

Your data is never used to train a model, ours or anyone else’s.

SOC 2 + SOX ready

Rippling monitors us continuously. Three environments, separated roles, and a full audit trail on every promotion.

ASOP-aligned

Built to the CAS syllabus and ASOPs 23, 41 and 56, the standards your reviewer is held to.

Tesora Workbench

Specialist agents produce the work, read-only review agents check it, and every figure cites the page it came from.

Pricing & rater deployment

Build a rater by talking to the application, then test and refine it until the numbers are right. Every change is versioned and auditable, and when you are ready to deploy there is no IT, actuarial or underwriting queue to sit in. Bring what you already have: your Excel raters all come across.

Reserving & Close

Most of the quarterly close, done. Triangles squared on ingest, methods side by side, and an IBNR with every value traced back to source.

a month, back

Per actuary, measured on our own workflows.

the output per actuary

Same headcount, same standards, twice the work through the door.

traceable to source

Every value Tesora populates cites the cell or page it came from.

How is this different from Claude, GPT or Copilot?

Those are models, and we route to them. The difference is everything around them: your data wired in once, every figure citing the page it came from, and the same question answered the same way twice. Plenty of answers happen to be right. Your auditor is asking a narrower question than that.

Can we import our existing rating models?

Yes, and it is usually the first thing we do. Send us the Excel rater and we rebuild it: the formulas get parsed and the factor tables get picked out. What comes back is a versioned model that runs your existing logic exactly as it runs today. It then has to pass the test suite before anybody can deploy it, so you can check it against your own numbers before you trust it with anything.

What integrations do you support?

Microsoft 365 including Excel and SharePoint, Databricks, Snowflake and the major cloud storage all connect directly today. Policy and claims systems are the ones that vary: every carrier has bent theirs into a shape nobody else has, so we map yours during onboarding rather than pretending there is a switch to flip. That mapping is part of onboarding and not a separate bill. Everything else goes through the REST API, and on a system we have not seen before the honest answer is that we will tell you once we have looked at it.

How does role separation work for SOX?

There are three environments and they are separate: actuaries build in dev, QA validates in test, and underwriters use prod. An actuary cannot push their own work to prod, which is the control your auditor is really asking about. Getting there means somebody else promoted it, and the trail shows who did it and when. Access follows the same split, so nobody holds two of those roles at once by accident.

Do you train on our data?

No. We hold zero data retention agreements with every model provider we route to, so your data is not kept by them and is not used to train anything, ours or theirs. If your security team wants those agreements in writing before you sign, ask and we will send them.

Is confidential data protected?

Yes. You get a single-tenant environment of your own, running on hardware nobody else is on, and your IT team controls it rather than us. Nothing about your book sits in a pool with anybody else's. If your security team would rather it ran inside your own infrastructure entirely, we can deploy it there.

Extraction

Reads a SERFF filing, an Excel rater or a script, and cites every value to the cell it came from.

Pricing

Trends and on-levels your own history, then fits severity and frequency separately.

Rating

Turns the selected rates into a tested rater you can call from underwriting.

Reserving

Squares the triangles, runs the methods side by side, and traces the IBNR back to source.