> For the complete documentation index, see [llms.txt](https://docs.connect.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.connect.ai/architecture.md).

# Architecture

**CNCT Architecture Overview**

CNCT is building a decentralized compute network that enables distributed AI inference and data processing through a globally coordinated network of worker nodes. The platform democratizes access to AI computing while ensuring reliable execution and fair compensation for resource providers.

<figure><picture><source srcset="/files/7mFSGW9nVvlvfeo6Q2io" media="(prefers-color-scheme: dark)"><img src="/files/oWYzJxyndXpqKgYr6tOe" alt=""></picture><figcaption></figcaption></figure>

**Core Components**

**API Gateway & Authentication** The CNCT API Gateway serves as the primary entry point, managing service discovery and request routing while ensuring secure access through CNCT ID authentication. This provides a unified interface for clients to interact with all platform services.

**Platform API Services**

* **Job Dispatcher**: Handles task distribution and state management across the network. It intelligently routes compute jobs based on node capabilities and reliability scores.
* **Validator Service**: Ensures quality and accuracy of computed results through verification protocols and performance monitoring.
* **Rewards Service**: Manages fair attribution and distribution of points based on node contributions and task completion.

**Storage Layer** CNCT Storage provides a distributed solution for task results and datasets, enabling efficient data access and persistence across the network. It's designed to handle both temporary computation results and permanent dataset storage.

**Worker Node Network** The network consists of distributed nodes (1...N\*N) that provide compute resources. Each node:

* Maintains a reputation score based on reliability and performance
* Can perform various tasks: data collection, processing, and ML inference
* Communicates securely through E2E encryption
* Submits verifiable results with unique job hashes

**Client Integration** Clients interact with the network through SDKs and APIs, submitting job requests and receiving model inference results. The platform handles all complexity of distributed execution while maintaining security and reliability.

**Node Application Layer** CNCT nodes can run as desktop applications, browser extensions, or web clients, providing flexible deployment options. They include:

* Task processing capabilities for various compute workloads
* Ledger API integration for managing points and rewards
* Secure result submission and reward claiming

**Marketplace & Developer Tools**

* **CNCT Store**: Offers AI applications, datasets, and infrastructure resources
* **Developer Portal**: Provides tools for dataset collection, application deployment, and API access
* **Point System**: Enables fair compensation through earned and spent points

This architecture ensures secure, reliable, and incentivized distributed computing while maintaining simplicity for end users and developers.