Zeroset Raises $5.2M for Enterprise AI Memory
Zeroset has emerged from stealth with $5.2M to build a continuously updated context layer for enterprise AI agents.
Enterprise agents often fail for a simple reason: retrieving documents is not the same as understanding what is happening inside a company.
What happened
Zeroset emerged from stealth with a $5.2 million pre-seed round. Its first product, Nebula, connects to enterprise tools and builds a continuously updated representation of workflows, decisions and organisational state.
The goal is to give AI agents more durable context about how a company actually operates rather than forcing each agent to reconstruct that information independently.
Why it matters
Many enterprise AI systems rely on search and retrieval. That works for finding documents, but real work depends on changing state: what has been approved, which customer issue is unresolved, who owns the next step and what happened earlier.
A shared context layer could make agents more reliable by grounding their actions in current organisational information.
The bigger picture
As enterprises deploy multiple agents, memory and shared state are becoming infrastructure problems in their own right.
Zeroset is betting that companies will need an internal world model sitting between their software systems and the agents acting on top of them. The round is small, but the architectural thesis is significant.