Skip to main content

AI agents (archetype)

AI agents (archetype)

Archetype pages on DecAIHub describe common project blueprints — recurring patterns of how crypto AI projects are built, what they offer, and what role AI plays in their product. The "AI agents" archetype covers projects whose primary product is building, deploying, or operating AI agents: autonomous software that perceives data, reasons about it, and takes actions without continuous human instruction.

In short: in this archetype the agent IS the product — not a feature bolted onto something else.

What defines this archetype

A project fits the AI agents archetype when its core offering revolves around agents rather than using agents as a secondary feature. The agent — its creation, deployment, operation, or monetization — is the main reason the project exists. This separates the archetype from projects that happen to use an agent internally (e.g., a DeFi protocol with an AI-assisted rebalancer) but whose primary product is something else. In the AI agents archetype, removing the agent removes the product.

Archetypes are descriptive, not endorsements. They help readers compare surface-similar roadmaps and ask consistent questions about evidence, safety, and architecture. Two projects in the same archetype can differ sharply in maturity and verifiable AI depth.

Common patterns

Within this archetype, several distinct product patterns appear repeatedly across the DecAIHub registry.

Agent frameworks

Build-your-own-agent toolkits. These projects provide SDKs, libraries, and developer tools for creating custom AI agents. They typically include pre-built modules for memory, planning, tool use, and chain interaction. The product is the framework itself — the agents built on it belong to third-party developers.

Typical AI components: agentic framework, agent SDK, AI agent platform (workflows/tools/KB).

Agent launchpads

Create, deploy, and tokenize agents. These platforms let users build an agent, assign it a token, and launch it on-chain — often with no-code or low-code interfaces. The pattern combines agent creation with token issuance in a single flow. This is one of the most common use cases in the category: AI agent creation + launch.

Typical use cases: AI agent creation + launch, build + deploy AI agents, no-code agent deployment.

Agent marketplaces

Discover and use pre-built agents. Marketplace projects aggregate agents built by others, letting users browse, rent, or purchase agent services. The marketplace handles discovery, payment (usually in tokens), and sometimes reputation. The AI belongs to individual agent creators, not the marketplace itself.

Typical AI components: agent marketplace, agent registry, agent discovery.

Agentic OS

Full operating systems for autonomous agents. These projects go beyond single frameworks and attempt to build a complete environment where agents can coordinate, share resources, use tools, and interact with multiple chains. They position themselves as the infrastructure layer for an "agent economy."

Typical AI components: agentic OS, multi-agent orchestration, agent swarm.

Vertical agents

Purpose-built agents for specific domains. Rather than providing a general platform, these projects build agents for a single vertical: DeFi trading agents, gaming NPCs, social media personalities, consumer assistants, or infrastructure management agents. The AI is narrowly focused and the value depends on how well the agent performs its specific task.

Typical use cases: AI bots/agents for crypto, AI-powered DeFi agent, autonomous trading agent, AI companion.

What unites these patterns

Despite different product shapes, all projects in this archetype share the same structural property: the agent is the product. The project does not merely use AI as one feature among many — the entire value proposition depends on the agent's existence, quality, and autonomy.

This means that evaluating projects in this archetype requires checking the agent itself: does it reason, or does it follow a fixed script? Does it use a real AI component (LLM, reinforcement learning, planning module), or is it a rule-based automation labeled as an "agent"?

Readers should also distinguish platform claims (“we enable agents”) from instance claims (“our agent did X on-chain”). Both can be valid, but they require different kinds of proof.

Where this archetype appears in DecAIHub cards

When reading a project card that belongs to this archetype, look for:

  • Category — typically AI/Agents, often combined with a domain tag (DeFi, Infra, Consumer, Gaming, Meme).
  • AI component — should name a concrete mechanism: agentic framework, agent marketplace, AI agent platform, agent swarm, multi-agent orchestration. Vague values are a warning sign.
  • Use case — common values: build + deploy AI agents, AI agent creation + launch, AI bots/agents for crypto, no-code agent builder. The use case should describe what the user does with the agent product.
  • ai_score — reflects how much verifiable evidence exists for the agent's AI capabilities. See DecAI Fit.

What this archetype does not prove

Being classified under the AI agents archetype means a project is positioned as agent-centric. The classification alone does not confirm that:

  • actual agents are running in production and performing autonomous tasks;
  • the agents use real AI rather than fixed scripts or simple API wrappers;
  • the platform has meaningful user or developer adoption;
  • the token is necessary for the agents to function;
  • the project's agents are safer, smarter, or more autonomous than alternatives.

Many agent projects are early-stage platforms with few agents actually deployed. The gap between vision and verifiable reality is often wide in this archetype — making the AI reality score and evidence tier especially important.

Do not confuse with

  • AI/Agents category — the top-level category label. The archetype page describes the common product pattern; the category page describes the classification system.
  • AI assistant archetype — assistants respond to user prompts but typically do not act autonomously. See assistant vs agent.
  • AI infra layer archetype — infrastructure projects provide tools and resources that agents may consume, but the infra itself is not an agent.
  • Automation — rule-based scripts that execute predefined logic without learning or reasoning. See Automation vs real AI.

Frequently asked questions

Is “archetype” the same as the AI/Agents category?

No. The AI/Agents category is a top-level label in the taxonomy. An archetype explains recurring product shapes inside and around that label (frameworks, marketplaces, vertical agents). Use archetypes for narrative pattern-matching; use categories and AI component values for structured comparison.

Why do two agent projects with similar landing pages score differently on ai_score?

ai_score reflects verifiable evidence for AI claims — documentation, code, demos, measurable behavior — not marketing similarity. Two agent-centric visions can diverge sharply once Tier-1 evidence is applied.

Should readers expect every agent marketplace to host proprietary model training?

Not necessarily. Marketplaces often focus on discovery, payments, and interfaces while creators supply models and policies. Evaluate each layer separately: platform infra, creator models, and on-chain execution evidence.

External references

See also

No comments to display

Back to top