Cumulative Computing
Traditional infrastructure preserves data. Cumulative Computing infrastructure preserves work.
Computational systems produce artifacts representing completed work. When those artifacts persist, the work persists. When they disappear, the work must be performed again.
Cumulative Computing is the paradigm in which systems are built to preserve and accumulate that work — rather than repeatedly destroying and recreating it.
The Core Idea
The Manifesto
A concise introduction to the Cumulative Computing paradigm and why it matters.
Read Cumulative Computing →Artifact Graphs
Computational results form dependency graphs that represent the structure of work.
Explore the framework →Computational Work Conservation
Completed computational work persists only through the artifacts produced by that work.
Read the principle →AAA Framework
Seven technical notes developing the Agent Artifact Availability framework from first principles.
Start with Paper 01 →Interested in this work? Researchers, developers, investors, and builders are welcome.
Get Involved →Key Concepts
- Artifact — The durable output of a completed computational process
- Artifact Graph — The directed graph of artifacts and their derivation relationships
- Artifact Identity — A stable identifier derived from the computation and inputs that produced an artifact
- Artifact Availability — The property that artifacts remain retrievable, verifiable, and reusable across agents, workflows, and time
- Computational Work — The accumulated structure of computation encoded in an artifact graph
- Computational Work Conservation — Completed computational work persists only through the artifacts produced by that work
- Cumulative Computing — The paradigm in which systems accumulate computational work over time rather than repeatedly destroying and recreating it