Luminos

LuminosAI (Luminos, Inc.) sells a software platform for AI governance and testing that evaluates classical machine learning, generative AI and agentic AI systems against legal, regulatory and reputational risks. Evaluations are built by the company's legal engineers and mapped to specific legal provisions and standards, such as FTC Act Section 5, CCPA/CPRA and unauthorized-practice-of-law exposure, and can be run without access to the underlying model. Results are returned as plain-English, audit-ready "report card" documentation intended for legal, privacy, compliance and board audiences, and customers can assert attorney-client privilege over testing results.

The platform's stated workflow has three steps: selecting or building an eval, connecting the target system's data either pre-deployment or in real time, and generating documentation. Deployed systems are re-evaluated continuously, with re-tests triggered by model updates, behavioral drift, or new deployment contexts, and findings routed to a human review queue when they warrant escalation. Additional platform components include automated documentation, ready-made review workflows and decision frameworks, a law library, and compliance reporting for existing and emerging AI laws. In February 2026 the company released Lighthouse, a feature that produces a consolidated risk "heatmap" for generative and agentic AI systems using structured sets of provisions the company calls Constitutions, each mapped to specific legal liabilities.

An earlier product line described in May 2024 comprised six governance products spanning automated reporting, bias calculation, legal privilege management and risk monitoring, developed over the preceding four years. At that time a generative-AI product called Judicial AI, which built a "legal constitution" used to train an AI "judge" that evaluates model outputs for legal compliance, was in prototype testing with a small number of customers.

Founding story

Luminos.AI was created out of Luminos.Law, a boutique law firm co-founded by Andrew Burt that focused solely on AI risks and was staffed by lawyers and data scientists. The firm had been helping clients build governance programs, auditing frameworks and red-teaming generative AI systems, and had developed custom software for that work. Because the firm was small and law firms cannot accept outside capital, the team decided to spin the software out as a separate company with its own team and outside investment. Members of the founding team had previously helped build Immuta, a data governance platform with origins in the US intelligence community.

Business model

Software platform sold to enterprises, with evaluations custom-built and run by the Luminos team rather than requiring a large customer setup effort; the company states there is no large setup fee. A partnership with law firm ZwillGen gives customers access to outside legal counsel directly within the platform when a finding requires legal action.

Traction

Six products offered as of May 2024, developed over the prior four years, with Judicial AI in prototype testing with a select group of customers at that time. The company states 90% of AI approvals are automated on its platform, and that early beta customers of Lighthouse reduced AI approval timelines from weeks to minutes, identifying risks in customer-facing chatbots and internal productivity tools. Testimonials come from Red Ventures.

Latest developments

On 24 February 2026 LuminosAI announced Lighthouse, described as enabling a 360-degree review of legal liability risk in generative and agentic AI systems, alongside the close of a funding round led by M13 with participation from Bloomberg Beta, Hawktail, AME Cloud Ventures, Crosscourt, Octave, Great Oaks and Fundrise. M13 partner Win Chevapravatdumrong joined the LuminosAI board.