Valency joins Lawrence Berkeley National Laboratory’s HERALD project, giving AI agents a first pass through a vast archive of nuclear power research so human experts can focus on the judgment calls that matter.
BERKELEY, Calif., July 22, 2026 — Valency, provider of foundational infrastructure for human-centered, AI-accelerated science, today announced its selection for the U.S. Department of Energy’s Genesis Mission. Valency joins Genesis as a funded partner in HERALD, a collaboration led by senior scientists Daniela Ushizima and Peter Nugent at Lawrence Berkeley National Laboratory, to use special-purpose AI agents as a tool to empower human security experts to review the vast archive of nuclear power research.
“Most document systems were built to store files and help people find them – not for AI agents that need to read, reason across, and act on huge collections of documents alongside human experts,” said Joshua Bloom, CEO of Valency. “That’s the infrastructure we’ve been building, and HERALD is a critical proving ground. We couldn’t have asked for a more meaningful scientific mission and partnership.”
“To advance nuclear power, we must unlock invaluable research trapped behind security reviews. HERALD will build AI to safely vet this trove of legacy data while maintaining the strict, auditable proof required for compliance,” said Dr. Daniela Ushizima, Berkeley Lab Senior Scientist and Principal Investigator of the Genesis Mission’s HERALD project in collaboration with Valency.
“The Valency Hub gives us something no research prototype offers: a production platform where every document, every AI verdict, and every reviewer decision is tracked in an auditable chain of custody. That’s what turns a promising classifier into infrastructure the DOE complex can actually trust and deploy,” said Peter Nugent, Division Deputy for Science for Berkeley Lab’s Applied Math and Computational Research Division and Systems Integration Lead for HERALD.
As part of the award, Valency powers the AI agent infrastructure behind HERALD: a system built to support hundreds of millions of scientific documents, many AI agents working at once, and a large group of human reviewers, all with the data provenance and controls that this work demands. Valency’s infrastructure speaks the Model Context Protocol (MCP) natively, giving HERALD’s agents a direct, structured way to work across the archive. Experts remain in charge of every decision that matters; the difference is they can spend their time on the documents that genuinely need their judgment.
About Valency
Valency is the infrastructure that keeps AI working from the real record, not a plausible-sounding approximation of it. Built to support hundreds of millions of scientific documents and the AI agents that read, reason across, and act on them, Valency provides researchers and their AI systems a grounded, current view of the record they depend on. Learn more at valency.io.
Media Contact
Matthew Work — Head of GTM
[email protected] • 415-412-4286
SOURCE Valency Systems Inc
