# Decentralized AI Network

Service providers (SPs) within the Griffin AI network serve a dual role as node operators, running the Griffin protocol software. Their primary function is to act as conduits between off-chain AI services and on-chain activities, facilitating seamless integration between these environments.

## SP responsibilities and behavior

* Each SP operating a node delivers specific services, which can range from:
  * General inference from hosted language models
  * Complex, specialized AI-orchestrated workflows
  * Any other service the SP supports
* SPs receive service requests:
  * Directly from users, or
  * Through Client Providers (CPs)
* In response to requests, SPs:
  * Activate the appropriate on-chain or off-chain processes
  * Deliver results back to the requesting user or CP

{% hint style="info" %}
If an AI model provider or other service provider cannot or will not run a Griffin node, an independent SP can be established to bridge the Griffin network and that AI provider using off-chain mechanisms.
{% endhint %}

## Participation requirement

* To offer services within the Griffin AI network, an SP must operate a node.
* By operating nodes, SPs also contribute to the network’s distributed database—ensuring integrity, continuity, validation, and synchronization.

## Node types and network roles

{% columns %}
{% column %}

#### SP nodes

* Run the Griffin protocol software.
* Provide and execute AI services and workflows.
* Bridge off-chain AI models to on-chain processes.
* Participate in distributed database maintenance.
  {% endcolumn %}

{% column %}

#### Guardian (full) nodes

* Operate full-nodes with oversight responsibilities.
* Monitor for compliance and potential malpractices.
* Play a key role in securing and validating network operations.
  {% endcolumn %}
  {% endcolumns %}

## Client Providers (CPs) and Light Nodes

* Client Providers (CPs) may operate:
  * Full-nodes, or
  * Light-nodes, depending on capacity and service needs.
* Light Nodes are optimized for lower resource requirements and enable broader participation:
  * Individual developers
  * AI agents
  * Smaller entities
  * CPs who prefer lower infrastructure overhead

Last updated 1 year ago

Links:

* <https://griffin-ai.gitbook.io/knowledge-base/overview-technical-architecture>
* <https://griffin-ai.gitbook.io/knowledge-base/decentralized-ai-network/network-design-and-functionality>