Imandra provides a cloud-scale automated reasoning platform that enhances large language models (LLMs) by converting their outputs into formal logic, enabling explicit and auditable reasoning. This technology eliminates inaccuracies, ensures compliance through formal verification, and allows for scalable logical inference in complex industrial applications.
Funding
$5M raised to dateRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.
GVFounders
Product
Problem
Large language models (LLMs) can produce inaccurate or untrustworthy outputs, hindering their adoption in industries where correctness and compliance are paramount. Current methods lack transparency and auditability, making it difficult to verify the reasoning behind LLM-generated results.
Solution
Imandra provides a reasoning platform that enhances LLMs by converting their outputs into formal logic, enabling explicit and auditable reasoning. This technology allows users to eliminate inaccuracies by making reasoning explainable, gain trust by formally verifying arguments, and scale reasoning to handle complex industrial applications. Imandra translates LLM outputs into the Imandra Modeling Language (IML), a formally defined subset of OCaml, and then uses automated reasoning to analyze the logic. The platform backs each answer with a sequence of logical steps that can be audited, providing transparency and verifiability.
Target Audience
Primary users are organizations in finance, government, and other industries where correctness, compliance, and transparency are critical, including financial firms, government agencies, and universities.
Features
- Converts LLM outputs into formal logic using the Imandra Modeling Language (IML)
- Employs automated logical reasoning to verify the correctness of LLM-generated results
- Provides a sequence of logical steps for each answer, enabling auditable reasoning
- Region Decomposition feature explains the behavior of complex software and algorithms, identifying edge cases
- Supports formal verification, optimization, constraint solving, symbolic reasoning, and rule synthesis
- Offers custom plug-ins to extend the reasoning engine for domain-specific applications
- Python library available for integration