Zeroset is an applied AI lab developing a “system of state” that provides a structured, queryable record of facts, timestamps, and provenance for automation workflows. By enabling continual learning and state‑conditioned reasoning, their Nebula platform offers a state layer for workflow automation and includes Atlas, the first benchmark suite for measuring stateful workflow performance.
Funding
Funding not disclosed
Founders
Product
Problem
Current automation systems lack a unified, queryable representation of factual state, making it difficult for models to know what is true, when it became true, and why. This gap prevents continual learning, state‑conditioned reasoning, and proactive behavior, limiting reliability and scalability of workflow automation.
Solution
Zeroset builds a structured state layer that records facts, timestamps, and causal context, enabling AI models to access up-to-date, queryable knowledge about the world they operate in. Their Nebula platform provides this state layer as a service, allowing downstream automation workflows to retrieve and update state information in real time. By integrating Nebula, models can perform continual learning, adjust reasoning based on current conditions, and act proactively rather than reactively. Nebula also includes Atlas, the first benchmark suite designed to evaluate performance of stateful workflows, giving developers measurable metrics for improvement. The platform is delivered via APIs that can be incorporated into existing automation pipelines, facilitating more reliable and adaptable process automation.
Target Audience
Primary customers are enterprises and developers building complex workflow automation systems that require reliable, up-to-date state information, such as AI‑driven process orchestration platforms and intelligent RPA solutions.
Features
- Structured, queryable state store capturing facts, timestamps, and causal provenance
- Real-time API for reading and updating state, enabling dynamic workflow adjustments
- Support for continual learning and state‑conditioned reasoning in downstream models
- Proactive behavior capabilities through access to up-to-date contextual information
- Atlas benchmark suite for quantifying stateful workflow performance across tasks
- Cloud‑native deployment with scalability for enterprise‑level automation workloads