Position paper explores the need for informational boundaries in long-horizon agents, suggesting new architectures.
This record archives the canonical version of the position paper and founder whitepaper introducing Agent Concretization, Schema Sandbox, and persistent Agent IP. The work is motivated by early engineering observations from psi.run, a newly launched persistent agent platform. It argues that long-horizon agents require not only memory, but active informational boundaries that regulate attention, action space, identity continuity, and computational resource use. This version should be read as a position paper and conceptual architecture whitepaper rather than a fully validated empirical study. The associated OSF/ECSarXiv preprint and supplemental materials provide additional project context.
No takes yet. Share an insight, caveat, or question.
LIU et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: