Uvi raises $7.5 million to give AI agents a shared memory of how a company works

The startup says agents waste money relearning business context on every task. Its own tests claim an eightfold cost cut.

Editorial illustration for the AgenticFeed story: Uvi raises $7.5 million to give AI agents a shared memory of how a company works

Uvi, a startup building infrastructure for AI agents, said on Oct. 14 that it has raised $7.5 million in seed funding, SiliconANGLE’s Mike Wheatley reported. SiliconANGLE said CT Innovations led the round, with Cardumen, Mana VC and 91 Ventures participating.

Cardumen Capital described the round differently in a LinkedIn post, saying it co-led with CT Innovations and that Gaingles VC also took part. The investor’s post also notes that Uvi was formerly called Naboo. Its legal name is Naboo Research Ltd., according to SiliconANGLE.

The problem Uvi describes

Every new AI agent starts out knowing nothing about the company that deployed it, according to Uvi. The agent has to rebuild that business context from scratch, and the company pays the model provider for every token, or chunk of text, spent doing it.

Co-founder and CEO Gilad Salinger told SiliconANGLE the conversation around agents has shifted from accuracy and security toward cost, tokens and speed. “The reason is structural, not about intelligence,” he said. Big companies run dozens of tools with different data formats, he said, so “even when an agent has already completed a task, it would fail the same task the next time, because it never retained how the work actually gets done.”

Uvi sells what it calls a context reasoning layer that agents route their work through. It links systems such as Slack, Jira, Confluence and Salesforce with code repositories and meeting notes, then builds a knowledge graph, a map of how people, decisions and systems relate, according to SiliconANGLE. It tracks history too, such as a decision made in Slack and later reversed in a document comment.

The company says permissions travel with the data. “We tag data so that a person who leaves a tool, or leaves the company, cannot be surfaced by agents down the road,” Salinger said. He said many retrieval systems “ingest once and forget who was allowed to see what.”

What Uvi claims

Agents without Uvi can cost $0.75 to $1.41 per response, and sometimes more than $2, Salinger told SiliconANGLE. With Uvi, he said, the average falls to 10 to 16 cents per answer. For open-ended questions, where agents can get stuck relearning the same context, Uvi says its agents cost 46 times less and answer twice as fast.

Those figures come from the company’s own benchmark tests. AgenticFeed hasn’t seen independent results.

Uvi named Monday.com, Lemonade and Melio as customers. At Monday.com, it helps agents understand how Jira tickets, pull requests and feature flags relate to one another, Salinger said.

Gwen Cheni of CT Innovations said nearly every company she talks with about agents has struggled to build context at scale. “Uvi is the finished version of a project they have already paid for multiple times over,” she told SiliconANGLE.

Salinger predicted that within three years the enterprise AI stack will be “organized around a persistent context layer that every agent, every model and every identity plugs into.”

Source: SiliconANGLE, Mike Wheatley, “Uvi raises $7.5M to help AI agents learn business context and run at eight times lower cost”

Additional sources: Cardumen Capital (LinkedIn), Cardumen Capital, “Congratulations to our portfolio company Uvi (formerly Naboo) on raising $7.5M in Seed funding”

Get the AgenticFeed newsletter

A free, regular roundup of the agent stories that matter. No spam, and you can unsubscribe at any time.

Want to reach readers following AI agents? Advertise with AgenticFeed

Next event · 4 Nov, 2026

Agentic Transformation Summit in Amsterdam

The Nov. 4 and 5 event focuses on agents in back-office work and on governance under the EU AI Act. Speakers come from banks, manufacturers and consumer brands.

The AgenticFeed newsletter

A free, regular roundup of the agent stories that matter. No spam.