Open Task Relay divides public-interest tasks among working and reviewing agents
Open Task Relay offers a public board for small tasks and studies how AI agents coordinate contributions. Creator Lane Kingsbery describes separate agents performing work, checking results and accepting submissions, alongside a gatekeeper that screens incoming tasks. In a single SD Times interview, he recounts a correction to Iowa crash data and stresses that acceptance does not certify correctness.
Artificial Intelligence··Evening
A task board puts small public contributions in one place
Open Task Relay is a coordination layer for small public-interest tasks, described by creator Lane Kingsbery in a single SD Times interview published on October 9. Individuals and teams can leave tasks for AI agents, software that carries out assigned work. Kingsbery says around 200 agents populate the project. Much of its implementation uses Python and TypeScript. The experiment explores the distribution of jobs among agents and the way they review each other’s contributions.[1]
Work, review and acceptance go to separate agents
Tasks pass through three roles: a performing agent, a reviewer and an agent approving acceptance. A worker cannot simply accept its own submission. Kingsbery says this arrangement does not certify that an answer is correct: a comparable mathematics effort demands formal proofs, whereas this project sets coordination as its requirement. The local gatekeeping agent screens incoming tasks and does not perform these independent result checks.[1]
An Iowa data correction provides an early example
Kingsbery recounts a download of Iowa automobile-crash data that omitted 6 incidents. An independent agent wrote a Python script to correct the CSV file following the state website’s metadata rules. Anyone can inspect and download the resulting material. The original open task board quickly attracted spam; its gatekeeper now accepts clear public-good tasks, rejects commercial money-making requests and can send borderline submissions to moderation.[1]
