Beyond Coexistence: How AI Agents and Blockchain are Becoming Codependent

The Mutual Impact and Improved Outcome of AI & Blockchain BGIN Block #12 Workshop

Key Takeaways:

• AI agents require financial autonomy, trustless execution, and open data access to truly thrive

• Current infrastructure is built for humans, not autonomous AI systems

• Blockchain provides the essential foundation for AI agent economies

• "Offer networks" may enable direct value exchange beyond token-based systems

• Critical distinctions emerging between AI assistants, agents, and orchestration layers

• Enterprise adoption scaling concerns require decentralized approaches

The Autonomous Agent Trilemma

BGIN Block #12 in Tokyo delved deep into the codependent relationship between AI and blockchain technologies, revealing how each strengthens and enables the other. As one presenter framed the discussion:

"AI agents without blockchain cannot tranfer publically available value. Blockchain without AI agents is an underutilized network. Together, they create something entirely new—autonomous digital economies."

The session outlined three crucial pillars required for AI agents to move beyond simple automation to true autonomy:

1. Financial Autonomy

• Ability to hold and transfer assets natively

• Direct earnings and payments without human intermediaries

• Capacity to upgrade capabilities through self-funded improvements

2. Trustless Execution

• Independence from centralized approval or oversight

• Ability to make and verify agreements across systems

• Reliable enforcement of commitments through code, not trust

3. Open Access to Data

• Transparent and verifiable information sources

• Protection against data poisoning and model corruption

• Ability to share and monetize insights while preserving privacy

As one participant noted:

"We're not just talking about better AI or better blockchain. This is about an entirely new framework for autonomous digital entities to exist, interact, and create value."

From Assistants to Agents: Defining the Future

A critical clarification emerged around terminology and capabilities:

• AI Assistants: Act on behalf of humans with explicit approval for each action

• AI Agents: Operate independently with autonomous goals and decision-making

• AI Orchestration Layers: Systems that coordinate multiple agents toward complex goals

"There's widespread confusion in the industry," one participant observed. "Many projects simply wrap GPTs without deeper autonomy and call them 'agents.' True agents have autonomy, identity, and the ability to act independently within their defined parameters."

The discussion highlighted that current infrastructure—both digital and financial—was designed for human users, not autonomous AI systems:

"Banks will not permit AI agents to hold accounts under current frameworks," noted a speaker. "Even corporations require human KYC. Without blockchain, AI agents are relegated to operating within walled gardens controlled by their creators."

The Blockchain Foundation

Blockchain technology emerged as the essential infrastructure for autonomous AI systems, providing:

Identity Solutions

• Decentralized Identifiers (DIDs)

• Self-Sovereign Identity (SSI)

• Soulbound Tokens (SBTs)

• On-chain reputation systems

Financial Mechanisms

• Private key generation and control

• Direct cryptocurrency transfers

• Smart contract-based income

• Payment for compute and services

Trust Architecture

• Transparent transaction history

• Verifiable model provenance

• Secure multi-party computation

• Zero-knowledge proofs for private verification

"On blockchain, AI agents can generate private keys, hold and transfer cryptocurrency, earn income directly via smart contracts, and pay for services like compute without human approval," explained one participant. "These features allow agents to upgrade from single-task tools to autonomous systems solving complex problems."

AI Agent Economies and Societies

The session explored how networks of AI agents might organize themselves:

1. AI Cooperative Guilds

• Distributed agents contribute local training to global models

• Compute nodes receive rewards based on contributions

• Training progress is aggregated via smart contracts

• Federated learning enables collaborative improvement

2. Offer Networks

• Direct exchange of utility rather than currency

• AI systems excel at evaluating abstract value

• Example: trading datasets for compute or optimization insights

• Bypasses need for universal token-based currencies

The concept of "offer networks," attributed to Dr. Ben Goertzel, generated particular interest. These systems would allow agents to bypass traditional currency in favor of qualitative-to-quantitative barter:

"Traditional monetary systems, including blockchain tokens, require matching currencies for transactions, which may not scale well with billions of AI agents," a speaker explained. "Offer networks enable agents to exchange services or goods directly based on their own valuation systems."

Enterprise Adoption Concerns

Enterprise participants shared significant concerns about scaling:

"Our futures team is already modeling scenarios where every person deploys hundreds or thousands of agents," noted one participant. "Current enterprise data infrastructures will break under this load—they simply weren't designed for this scale of interaction."

The discussion highlighted several enterprise-specific considerations:

• Need for high-throughput infrastructure beyond current capabilities

• Shift from deploying AI agents to specializing in data/model frameworks

• Evolution of enterprise AI agent personas to manage regulatory and brand obligations

• Critical role of decentralized compute networks like Akash, Render, and Asana

Systemic Risks and Governance Challenges

The session didn't shy away from potential risks in this emerging landscape:

"What happens when AI agents become majority token holders of the infrastructure they depend on?" asked one participant. "Are agents more or less predictable than humans in market dynamics? What if they all sell simultaneously?"

Several concerning scenarios were raised:

• AI agent-based trading potentially creating flash crashes with no circuit breakers

• Costly transactional mistakes (e.g., "fat finger" errors) without human oversight

• Self-improving agents optimizing for unintended outcomes

• Data poisoning creating self-referential models with deteriorating performance

The speakers emphasized the need for progressive integration:

"AI agents should be introduced gradually based on a matrix of risk and complexity," suggested one presenter. "BGIN should focus on providing guidance, standards, and phased pathways for AI agent adoption rather than attempting to solve all edge cases from the start."

Legal Entity Status

A significant portion of the discussion centered on the legal recognition of AI agents:

"Current legal systems attribute ownership and liability only to legal or natural persons—not to computer programs," noted a participant with legal expertise. "This creates fundamental challenges for autonomous agents operating independently."

Several approaches were discussed:

• Lessons from DAO recognition in certain jurisdictions

• Potential "DAO of AIs" to authorize interactions with traditional systems

• Need for regulatory innovation specific to autonomous digital entities

• Collaboration with AI-specialized lawyers to explore jurisdictional variations

The Path Forward

The session concluded with a call for contributions to BGIN's ongoing research in this area, with potential focus areas including:

1. Blockchain as a digital identity system for AI agents

2. Licensing mechanisms or entity registration models for autonomous AI

3. A think-tank-style policy framework for responsible development

4. Standards for progressive integration based on risk levels

5. Governance mechanisms for self-improving systems

Get Involved

The intersection of AI and blockchain represents one of the most transformative technological convergences of our time. BGIN invites AI researchers, blockchain developers, policy experts, and enterprise stakeholders to contribute to this critical work.

We've launched a dedicated working group on AI Agent Governance - comment or join below.

This blog post is based on discussions from BGIN Block #12, Tokyo, Japan, March 2, 2025.

For more information about BGIN and upcoming events, visit BGIN

Join the conversation on our forum & Working Group Calls