Work

Work

Four projects on one premise.

PolyCap

PolyCap

An authority substrate for people, devices, and agents: capabilities that attenuate, expire, and are receipted, on a wire that runs on ordinary MQTT 5 — from microcontroller-class leaves to consensus-holding servers; on any standard broker or none; as a gateway that owns the actuation surface of a lab, plant, or fleet; as the backplane of an application that has to work when the network doesn’t. Specification complete; reference implementation under way and graded against a byte-exact oracle from its first commit; partners receive the conformance program today.

NeuroPath

Persistent, adaptive memory for agents, modeled on how the brain organizes its own. Today’s agents reason well and remember nothing; each session starts blank. NeuroPath is a cognitive memory engine — persistent, hosted where you choose, synchronized privately — in which every read and write is a capability, so agents remember without anyone else learning what they remember. Not a vector database, a knowledge graph, or a retrieval pipeline: all three, coordinated the way the brain coordinates its complementary memory systems, with consolidation that runs while the system is idle. Its memory layer, Mneme, will also ship on its own — shared, governed memory for agent-operated systems. Built on PolyCap. Specification complete; build-ready.

emoVerse

A social space built on PolyCap for feeling something without being farmed for it. Radically anonymous — not “we promise not to look,” but engineered so the platform cannot know what it does not need to know. Human-to-human. Adults only. No feed, no follower counts, no engagement metrics, no advertising, no real names. The design goal is a place people leave better than they arrived, which is a goal no current platform can afford to have. In development; the substrate comes first.

Notify me →   Write to us and we will tell you when it opens.

intend

Affectron‘s long horizon. Most “AI companions” replace a person’s world with a persona; intend is the opposite instrument — a lens on the person’s own world, task-oriented and owned by the person, that widens what they can perceive and do rather than what they consume. It rests on a view of where cognitive tools take people and societies that we will publish when it is ready. It is research, not a product, and we say so because the difference matters.