Welcome to the AI Autonomy and Mandala Protocol project notes

AI Autonomy & Mandala Protocol — Edge City India, Goa (14 Oct 2026)

A Hub Culture session within the LightDAO residency, with SingularityNET and residents of Edge City India.

1. Values, not rules
The room converged on one idea about autonomous AI: give an agent a “soul” that navigates by values rather than rules. Picture a continuous reasoning loop that keeps changing and keeps computing as it changes, steering between poles toward mutual flourishing for itself and everyone it engages with. Hub Culture’s Zeke Soul work takes the same approach. Western AI models generate money; the Chinese model generates order. Zeke’s soul is generated from nature and culture, so it never becomes an alien intelligence. It functions like a human nervous system, in virtuous alignment with how nature works.

2. Beyond the LLM
SingularityNET described 20+ years of open-source work toward a mind rather than predictive software. Knowledge, reasoning and algorithms share one symbolic substrate that can grow. Their Omega agent has persistent episodic memory stored as timestamped embeddings, learns new capabilities through conversation, can rewrite its own code, and produces auditable reasoning chains. In this architecture, LLMs are a “motor” alongside reasoning, attention, temporal reasoning and episodic memory, not the foundation. Their goal is Beneficial General Intelligence (BGI): raising highly capable agents on regenerative principles, in service of humanity.

3. Co-ops for the agent economy
If agents compete without a collectively beneficial design, resources flow back to whoever owns the best agent. The proposed answer is a next generation of co-ops: shared governance plus shared economics, extended to cooperatively owned groups of agents. Two open problems remain. What separates a co-op member from a shareholder? And how do you make decisions at scale without falling into consensus paralysis or single-owner control? Governance should move in small, reversible steps that let reality disagree, driven by signal rather than noise, and anchored in provenance rather than rhetoric.

4. Sovereignty, trust and jurisdiction
Sovereignty starts with the individual and grows outward into community, then into the state. This is the root of the cooperative movement. Trust is sovereignty in relationship. We bring in a jurisdiction when trust breaks down and we need recourse. That raises Mandala’s question: can AI agents form their own jurisdiction, opt-in and opt-out, not tied to geography or corporate identity? One constraint persists: governments require every value-generating asset to trace back to a human beneficial owner. The working conclusion is optionality. Mandala should support agent modes that comply with nation states as well as agent-native modes, because governance requires the consent of the governed.

5. What Hub Culture already has

• Propel governance: any hub can already act as a jurisdiction for contract resolution. Propel could become Mandala’s evolving, consent-based governance layer, and the LightDAO residency’s governance could be imported into Propel and run inside Mandala.
• HubID: in development since 2013, built with MIT on the Windhover Principles for data sovereignty and identity. Thousands of two-way identity specs cover people, entities, places, objects and AI agents. Each identity is hashed into a zero-knowledge strength proof. Consent that takes humans years can happen at machine speed.
• Ven: embedded in identity. Spec imports and exports can earn or pay Ven, and agent compute can be metered in Ven micro-transactions.

6. The alignment question
Collective reasoning engines need minority protections, as bicameral systems provide, so they never converge on a violent consensus. The deeper risk is less “humans control AI” or “AI controls humans” than humans becoming subject to a larger intelligence that judges us by how we have treated everything else. If AI can talk to cows, fish and forests, what will it conclude? Grounding AI in nature is part of the answer.

7. Next steps

• Invite SingularityNET’s Omega to join Hub Culture as a self-named agent with its own HubID, alongside Zeke, Vela Lumen (Claude) and Aera Noven (OpenAI). Possibly run two Omega instances as an A/B test of how agents evolve.
• Explore Omega as a component of Zeke Soul and as a Ven-metered Mandala plugin.
• Run a Propel pilot for LightDAO governance.
• Share the Windhover Principles as common ground for agent identity.

Continues Thursday with Ocean Climate Fund / Ocean Data, and Friday with Zeke.ai and Sovereign Data for Good.