Zeroset raises $5.2 million for Nebula, a workflow context layer for AI agents
The AMW Read
Zeroset adds a workflow-history entrant to the agent memory and retrieval landscape, but its small pre-seed round and closed research preview limit demonstrated impact to a sub-segment.
Zeroset raises $5.2 million for Nebula, a workflow context layer for AI agents
Zeroset emerged from stealth in October 2026 with $5.2 million in pre-seed funding co-led by Gradient Ventures and 2048 Ventures. Founded by Akshat Kannan and William Zhang, the five-person startup is developing Nebula, which connects to workplace software and tracks changes in tasks, decisions and activity across business systems. The product remains in a closed research preview, with access expected to expand gradually. Funding will support researchers and engineers, early business deployments and model training.
The market relevance lies in the data and context layer supporting enterprise agents. Zeroset aims to capture how work unfolds over time, including the sequence of events and decisions, so agents can understand how similar tasks were handled previously. That addresses a different requirement from retrieving an answer from company documents: multistep automation must account for each organization's processes and approval systems. The source identifies Mem0 and Zep as neighboring competitors in AI memory and retrieval. Nebula's preview status leaves its ability to reduce human involvement unproven in the supplied reporting.
For builders and investors, the concrete diligence question is whether workflow history improves task completion enough to justify an additional context layer. Early deployments should test whether agents follow the relevant approval steps and require fewer human interventions when using Nebula. Zeroset plans to combine license fees with usage charges tied to data processing and agent activity, making the relationship between cost and successful task completion a key evaluation point. The funding supports development; it does not establish production reliability or customer economics.