About Lorena Labs

We build Tarski, a platform for expert systems whose rules are written by coding agents, whose input is read by LLMs, and whose decisions people can check.

Why we built Tarski

Much of the expertise in a business is a set of rules applied to cases: when to approve, when to escalate, when to refuse. Expert systems tried to encode those rules in the 1970s and 1980s. They worked in narrow domains and were expensive to build and maintain.

Two things have changed since then. Coding agents can write and maintain large rule sets, and LLMs can read the unstructured input that rule systems never handled well. Tarski puts the two together. LLMs propose facts and actions, your rules decide which to accept, and every decision can be traced back to its inputs and replayed.

The engine is built on established results from logic, including fixed-point semantics, locality, and the guarded fragment. They are what let Tarski compute state deterministically, check invariants before any action runs, and explain each conclusion.

the name

Tarski is named after Alfred Tarski, the logician who gave the formal definition of truth in a model. The Knaster–Tarski fixed-point theorem underlies how the engine evaluates rules.

Careers

We are a small team working on the engine, the product, and developer experience. We are not hiring right now. When we open a role, it will be listed here first.

You can still tell us what you would want to work on, and we will keep it on file.

Tarski is in private beta.