Uvi Secures $7.5M Seed to Give Enterprise AI Agents the Missing Context Behind Business Decisions

The initial rush to deploy enterprise AI agents has hit a wall: while models can search company documents, they struggle to understand the scattered, undocumented context behind business decisions. To solve this, Tel Aviv-based startup Uvi has emerged from stealth with a $7.5 million Seed round led by CT Innovations and Cardumen Capital. The company is building an infrastructure layer that maps how internal entities relate, rather than just indexing text.

Traditional AI deployments rely on vector databases that match keywords but miss the sequence of events. Uvi replaces this with a "Decision Graph" that connects fragmented workflows—tracing a product decision from an initial Slack thread, through a Jira ticket, and into the final codebase. When an employee asks an AI model like Claude or GPT why a feature was delayed, the agent can follow this mapped reasoning instead of guessing based on isolated documents.

Unlike standard Retrieval-Augmented Generation (RAG) setups that often leak sensitive data, Uvi enforces existing access controls at the graph level, ensuring agents only retrieve information the user is authorized to see. Because the knowledge graph sits outside any single AI model, companies can swap between different foundation models without needing to retrain them on business logic. Early adopters include Lemonade and monday.com, with Uvi claiming its approach reduces query costs by up to eight times compared to custom-built agents.

The funding highlights a shift in enterprise AI from raw compute power to contextual accuracy. As organizations move toward autonomous agentic workflows, the bottleneck is no longer the intelligence of the foundation model, but its ability to navigate the messy reality of corporate data.