Legal AI vendor GenieAI claims 90% accuracy beating Claude Cowork, ChatGPT in self-run benchmark; no independent verification

GenieAI, a legal-contract AI agent marketed for drafting, reviewing and negotiating agreements, says on its site that "Genie achieved 90% legal accuracy, compared to 79.3% for Claude Cowork and 37.3% for ChatGPT," according to the company's published materials reviewed by AI Dispatch. The company does not publish the benchmark's methodology, test set, or an independent party that ran or verified the comparison on the page reviewed.

The company describes its product as "a purpose-built legal AI platform powered by a patent-pending architecture called Eidetic Intelligence," built for teams that "draft, review, edit, and negotiate agreements quickly and with confidence." It markets the tool for high-volume commercial work, saying customers can go "From EPCs to supply agreements. Automate high-volume contracts with confidence" — EPC referring to engineering, procurement and construction agreements.

GenieAI reports "We are ISO27001 certified, so your data is secure," and says "We do not train on your data. Genie improves independently. All data stored on Genie is private to your organisation." The company's site displays a running total of "300 K+ Contracts" processed on the platform and says it is "Trusted by 200,000+ businesses," without further detail on either figure.

Pricing starts at "Free From $0 per month," with a paid tier "Pro From $75 per month" and an Enterprise tier that adds an "SLA with technical support" and a "Dedicated CS Manager & onboarding," the company says. The product supports "MCP connectors, API, and native integrations with email, CRM, cloud storage, and eSign."

What GenieAI's page does not publish: any audit-trail mechanism for agent actions, insurance or indemnification terms, a liability cap, documented incidents, or independent safety or accuracy evaluation of a product performing legal contract review and negotiation — work that carries real professional and financial consequences if the AI gets a clause wrong. The 90%-accuracy comparison against two named competitor products is the company's own figure; AI Dispatch found no independent benchmark, replication, or third-party audit of it on the site reviewed.