Zeroset: Funding, Valuation and AI Agent Memory
Zeroset is an applied AI research startup developing the memory and state infrastructure needed for autonomous agents to work reliably inside enterprises. Its first product, Nebula, connects workplace systems and tracks how tasks, decisions, handoffs and business information change over time. Instead of reconstructing the state of a workflow from disconnected documents every time an agent runs, Zeroset aims to give it a structured, current understanding of what has happened and what needs to happen next.
Visit Official Website ↗Basic Information
| Field | Details |
|---|---|
| Company Name | Zeroset |
| Official Website | zeroset.com |
| Headquarters | San Francisco, California, United States |
| Founded | 2026, according to company and funding coverage |
| Industry | Artificial intelligence, enterprise software, AI memory and workflow automation |
| Co-Founders | Akshat Kannan and William Zhang |
| Core Focus | Building a persistent memory and state layer for autonomous AI agents |
| Primary Product | Nebula |
| Funding Stage | Pre-seed |
| Latest Funding Round | $5.2 million pre-seed, announced October 6, 2026 |
| Total Disclosed Funding | $5.2 million |
| Lead Investors | Gradient Ventures and 2048 Ventures, co-leading the round |
| Other Investors | Leblon Capital and other participants |
| Valuation | Not publicly disclosed |
| Product Availability | Nebula was announced in a closed research preview at launch. |
What Zeroset Does
Enterprise Memory for AI Agents
Zeroset aims to give agents a structured understanding of an organization’s work. It connects information from workplace systems and represents how decisions, tasks and relationships evolve rather than treating each document as an isolated source.
Workflow State Tracking
Nebula tracks the current state of work, including what has happened, what has changed, what remains open and how different activities connect. Agents can use this information to avoid repeating steps or acting on stale context.
Context for Autonomous Decisions
AI models can generate plans but still struggle to understand the specific history and operating rules of a company. Zeroset is developing infrastructure that gives agents more relevant context for deciding what to do next.
Key Products and Capabilities
Nebula State Layer
Nebula is Zeroset’s first product. It builds a structured, queryable representation of enterprise activity so agents can retrieve the current state of a workflow without reconstructing it from raw records on every run.
Workflow Traces
The platform captures handoffs, actions, messages, decisions and other traces across connected systems. These records help explain how work moved from one stage to another.
Entity Resolution
Nebula connects references to the same people, tasks, projects and other entities across different applications. This helps agents reason about related activities even when information is spread across multiple tools.
Temporal Context
Business information changes over time. Nebula is designed to preserve changes and timing so agents can distinguish current information from earlier states and decisions.
Provenance and Contradictions
The platform aims to preserve where information came from and surface conflicting records. This helps agents assess evidence instead of treating every stored statement as equally reliable.
Repeatable Workflow Playbooks
By examining how work is completed across systems, Nebula can turn successful patterns into repeatable playbooks that agents can follow and improve over time.
How Zeroset Approaches AI Memory
State Rather Than Search Alone
Traditional retrieval systems find information similar to a query. Zeroset focuses on representing the state of a business, including relationships and changes, so an agent can understand the situation it is operating in.
Persistent Operational Context
Long-running agents need continuity between executions. Nebula is designed to retain relevant workflow history so an agent does not have to reconstruct the entire situation from scratch each time.
Lower Context Overhead
By retrieving the smallest relevant current state for an action, the platform aims to reduce unnecessary context passed into models and limit repeated processing of large amounts of information.
Cross-System Understanding
Enterprise work spans email, collaboration tools, tickets, code repositories and approval systems. Zeroset connects these traces to help agents understand the complete workflow rather than a single application.
Learning from How Work Happens
The company envisions using structured workflow traces to develop enterprise-specific models of how a business operates and changes, moving beyond static document retrieval toward operational understanding.
Traceable Decisions
Connecting decisions to the underlying events and records can help agents explain why a workflow is in its current state and what information supports the next action.
Nebula Integrations and Use Cases
Microsoft 365 and Outlook
Connecting workplace communication and documents can help agents follow decisions, responsibilities and ongoing tasks across email and collaboration workflows.
Teams and SharePoint
Information shared in team conversations and document repositories can provide context about project status, approvals, handoffs and the history behind decisions.
GitHub and Engineering Workflows
Engineering agents can benefit from understanding how code changes, issues, reviews and decisions relate to the wider development process.
Enterprise Operations
Agents handling multi-step operational tasks can use workflow state to identify outstanding actions, dependencies and the next appropriate step.
Financial and Business Research
Research workflows often involve information from multiple sources and repeated updates. Structured context can help agents maintain continuity as evidence and decisions change.
Multi-Step Automation
By tracking handoffs and unresolved work, Nebula aims to help agents complete longer workflows with fewer repeated steps and less need for human intervention.
Zeroset at a Glance
Funding Raised
Pre-seed financing co-led by Gradient Ventures and 2048 Ventures.
Core Product
A memory and state layer designed to help AI agents understand enterprise workflows over time.
Valuation
No verified public company valuation was identified in the sources reviewed.
Who Can Use Zeroset?
Enterprise AI Teams
Teams building agents that work across company systems can use a structured state layer to maintain context between actions and reduce repeated reconstruction of workflow history.
Workflow Automation Developers
Developers creating agents for multi-step business processes can use workflow traces and current-state information to help agents understand dependencies, handoffs and outstanding tasks.
AI-Native Businesses
Companies adopting autonomous agents across engineering, operations and research can explore enterprise memory infrastructure to make their systems more consistent as work evolves.
Funding Journey
Zeroset developed its initial product around the idea that enterprise agents need a persistent representation of business state, not only document retrieval.
Gradient Ventures and 2048 Ventures co-led the round, with participation from Leblon Capital and others.
Nebula was introduced in a closed research preview. The announced funding is intended to support product development, hiring, model training and early enterprise deployments.