The architectural decisions in IsoKron — schema-first typed graphs, hash-chained audit, BYO-fleet on consumer hardware, the security review pass before every commit — came out of months of structured deliberation with frontier-model councils. The decisions are public because the reasoning should be too.

Papers landing in this section will cover things like why our service is meaningfully cheaper per token than the alternatives, why we ship structured outputs instead of chat history, why we chose hash chains over signed logs, and the comparative analyses we ran on memory layers, agent frameworks, and hosted coding agents.

Papers coming soon. If you want early access to drafts as they land, dial in.

chain_height: pending·verified pending