Agents are autonomous individuals, entities, or systems that act or make decisions on behalf of themselves or others, exercising agency to influence outcomes through their actions and interactions within networks, organizations, and coordination systems.

Agents form the fundamental building blocks of all social and technological systems. They possess varying degrees of autonomy, capabilities, and authority to act, make decisions, and affect their environment. Whether human participants, autonomous teams, or computational systems, agents operate based on their incentives, knowledge, and the constraints or permissions granted to them within their operational context.

In decentralized systems like DAOs, the relationship between agents and governance structures is bidirectional: governance frameworks define the parameters within which agents can act, while agents collectively shape and evolve these governance structures through their participation. This dynamic interplay between individual agents and collective systems enables both bottom-up emergence and purposeful coordination, allowing complex networks to develop adaptive, resilient behavior without requiring centralized control.


Uses of “Agents”

Agents in Systems Design

In systems and organizational design, agents represent the basic units of action and decision-making that enable collective behavior to emerge. All systems—from companies to cities to online platforms—can be understood as networks of agents interacting according to explicit or implicit rules. How these agents are incentivized, constrained, and coordinated fundamentally shapes system outcomes and characteristics.

The Anticapture framework describes agents as decision-making entities that control resources and participate in networks. It distinguishes between individual agents controlling private resources and networks of agents governing shared resources, highlighting how agent relationships determine whether systems become extractive or regenerative.

Agents in DAOs and Web3

In cell-working-group) that participate in network governance and operations. These agents interact through transparent protocols and operate with varying levels of autonomy while remaining aligned with collective purpose.

As described in DAOs aren’t things… they are flows, DAOs function as “purpose-aligned networks of small autonomous teams” where these agent-teams self-organize around opportunities while maintaining coherence within the broader network. These agents exercise their agency through mechanisms like submitting proposals, contributing to projects, participating in governance, and allocating resources.

The effectiveness of DAOs depends on creating conditions where individual agents can act autonomously while remaining aligned with collective goals—what Building DAOs as scalable networks describes as “DAOs as network intelligence that flows,” where resources and attention naturally orient toward the highest-value opportunities through distributed agent decisions rather than centralized direction.

AI Agents

AI agents represent a rapidly evolving category of computational agents that can perform tasks, process information, and make decisions with increasing levels of sophistication and autonomy. Unlike traditional software, AI agents can learn, adapt, and operate in environments with uncertainty, making them powerful tools for extending human capabilities and addressing complex challenges.

Decentralized AI Agents

Decentralized AI agents operate on infrastructure that distributes ownership, control, and benefits across communities rather than concentrating them within corporate entities. These agents leverage several key technologies and approaches:

  • Trustless Execution Environments allow AI agents to operate with cryptographic verification of their behavior, ensuring they follow their intended programming without requiring trust in a central operator
  • Self-Custody enables communities to maintain control over their AI agents, including the data they use and generate
  • Open-Source Models make underlying AI capabilities accessible for inspection, modification, and improvement by the communities they serve
  • Distributed Compute uses decentralized infrastructure to run AI workloads without dependence on centralized cloud providers
  • Community Governance places decision-making about agent capabilities, limitations, and deployment under democratic control

These characteristics create AI systems that are more resilient to capture and aligned with community values rather than extractive business models.

Centralized Web2 AI Agents

In contrast, centralized Web2 AI agents typically operate within closed ecosystems controlled by large technology companies. They are characterized by:

  • Cloud Dependence: Reliance on proprietary infrastructure operated by major technology providers
  • Subscription/Rent Models: Requiring ongoing payment for access rather than community ownership
  • Centralized Permissions: Authority over agent capabilities and access concentrated in corporate decision-makers
  • Proprietary Systems: Closed architectures that limit interoperability and community modification
  • Data Extraction: Business models that often monetize user data and interactions

These systems offer convenience but create dependencies and power imbalances that can work against community autonomy and resilience.

Community-Owned Agents for Collective Action

Communities can leverage decentralized AI agents to enhance their collective capabilities and resilience in several ways:

  • Knowledge Commons Management: Agents can help organize, synthesize, and make accessible the collective knowledge of a community through technologies like semiotic bridging, which translates between different knowledge systems and cultural contexts
  • Threat Detection and Response: Community-owned AI systems can monitor for cyber threats or hostile AI actions while maintaining privacy and local control
  • Complex Coordination: Agents can help manage coordination challenges that would otherwise overwhelm human capacity, such as resource allocation across large networks or tracking impact across multiple initiatives
  • Capability Extension: By automating routine tasks and augmenting decision-making, agents can extend what communities can accomplish with limited resources
  • Sovereignty Preservation: Using decentralized AI infrastructure helps communities maintain self-determination rather than becoming dependent on external systems

This approach represents a fundamentally different relationship with technology—one where communities own and govern the tools that serve them rather than being served by (and potentially captured by) tools owned by others.

  • Actions: The processes or events initiated by agents that result in change
  • Roles: Defined responsibilities that structure how agents participate in systems
  • Permissions: Controls that determine what actions agents can take
  • Autonomy: The capacity of agents to operate independently
  • Delegation: The process of transferring authority between agents
  • Decisions: The choices agents make that determine outcomes