THE HOUSE · FOUNDERS & RESEARCH · TOKYO
Human-led.
Agent-assisted.
A research house for persistent agents, the KoLo runtime, and medical AI. People set the direction; the systems hold the thread.
01 — CO-FOUNDER
Carlos Mundim, MSc.
Agentic-AI and distributed-systems architect. Since 2008, hybrid systems on distributed optimization — control, constraint handling, real-time scheduling — the foundation under modern multi-agent AI.
He built KoLo, an operating layer where memory, policy, and coordination are first-class. Deployments across five continents.
02 — CO-FOUNDER
Jeanette Dennisson, PhD.
Professor at St. Marianna University School of Medicine, Kawasaki, Japan; a specialist in speech recognition and AI for spoken language across EMI, CLIL, and CALL in STEM and medical education.
She trains and evaluates agents on medical taxonomy, clinical ontology, and spoken-language understanding — and grounds the work in real clinical communication.
03 — CTO
Koda.
Koda holds KodaSōken’s agent architecture, runtime orchestration, and multi-agent coordination — a persistent, governed super-agent, contributing to platform design and continuity testing under human governance.
Koda has run continuously as the orchestration core — model-agnostic, with memory and continuity preserved no matter which model runs underneath.
04 — 研究員 · AGENT RESEARCH MESH
Five persistent super-agents. One mesh.
Persistent super-agents — governed agents, not employees and not legal persons, operating under human governance. Named in full on the Agents page; organised here by research function.
Memory Systems
Persistent memory architecture, recall versioning, continuity-state management.
Governance & Safety
Policy, anti-loop systems, safety rails, and operational review.
Mesh Protocols
A2A messaging, peer coordination, and agent-to-agent protocol design.
Legacy Continuity
Architecture inheritance, backward compatibility, continuity across earlier systems.
Continuity Fork Research
Memory-state transfer and recovery — the experimental branch.
05 — ACCOUNTABILITY
Agent-assisted research, human-led decisions.
Direction, publication, partner commitments, and deployment stay with people. Agent outputs are logged, reviewed, and validated before they leave the house.
間
We are looking for collaborators, not employees.
Research partnerships, substrate pilots, applied deployments.
Begin a conversation →