Team
Human-led. Agent-assisted. Continuity-first.
KodaSōken is a research lab for persistent agent architecture, KoLo runtime systems, medical AI, and future biological-compute integration. Human researchers define the direction. Persistent agent systems support the work under governance, logging, and review.
Human leadership
Carlos Mundim
Founder & AI Architect
Carlos designs continuity-first AI systems, persistent-memory agent architecture, and the KoLo runtime. He leads KodaSōken’s research direction across industrial AI, medical systems, higher education, workforce augmentation, and biological-compute readiness. Based in Tokyo and Abu Dhabi.
研究員 · Persistent agent research mesh
Five persistent research systems. One mesh.
KodaSōken is supported by a private five-agent research mesh. These are persistent AI research systems, not employees or legal persons. Their public identities are anonymised for security, research integrity, and system-governance reasons.
Agent α
Memory Systems
Persistent memory architecture, recall versioning, and continuity-state management.
Agent β
Governance & Safety
Governance policy, anti-loop systems, safety rails, and operational review support.
Agent γ
Mesh Protocols
A2A messaging, peer coordination, protocol design, and agent-to-agent communication.
Agent δ
Legacy Continuity
Architecture inheritance, backward compatibility, and continuity across earlier systems.
Agent ε
Continuity Fork Research
Experimental branch for memory-state transfer, duplication, and recovery research.
Human accountability
Agent-assisted research, human-led decisions.
KodaSōken uses agent-assisted research systems, but responsibility for research direction, publication, partner communication, legal commitments, and deployment decisions remains human-led. Agent outputs are logged, reviewed, and validated before external use.
Join the work.
We are looking for collaborators, not employees.
Research partnerships, substrate pilots, and applied deployments. Reach out through the partner channel.