Tech: From Request to Result: How Users & Nodes Interact in Cortensor
Mining is just one part of Cortensor β but what happens when users send AI requests?
Letβs break down how sessions, task execution, and real-time AI inference work, all while supporting both Web2 & Web3 environments.
π Node Reputation & Ephemeral Nodes
Once nodes complete Cognitive tasks, they build Node Reputation over time.
π‘ High-reputation nodes become Ephemeral Nodes, meaning theyβre eligible to serve real-time user requests dynamically whenever a task is submitted.
π Session Creation: The AI Subscription Model
Before a user submits AI tasks, they create a session β similar to subscribing to an AI service. Sessions define:
β
Task configuration (accuracy, validation, cost)
β
Deposit of $COR Tokens
β
Task execution lifecycle
βοΈ Session Configuration
Users can customize their session:
β Higher validation levels = More miners & accuracy (but higher cost)
β Lower validation levels = Faster & cheaper inference
This functions like Service Level Agreements (SLAs) in cloud computing β providing flexibility for different AI needs.
π° $COR Token Deposits
Users deposit $COR tokens in advance. These tokens are deducted as tasks are processed β just like a pay-as-you-go model for AI inference.
β Ensures miners get paid fairly
β Keeps the network economically sustainable
π Router Node Assignment
After creating a session, a Router Node is assigned to handle all interactions between the user and the network.
π The Router Node:
β
Routes requests to miners
β
Ensures tasks are assigned efficiently
β
Bridges Web2 & Web3 AI task execution
π Submitting an AI Task
Once a session is active, users submit an AI request. Example: βSummarize this article.β
Hereβs how Cortensor processes it:
β Task metadata is stored on IPFS
β CID (content ID) is recorded on-chain
β Miners fetch the job and start processing
βοΈ Task Processing by Miner Nodes
The Router Node queues the request and assigns Ephemeral Nodes to process it.
β Miners run AI inference based on user configuration
β Results are validated via Proof of Inference (PoI)
β Only the best-quality AI responses are accepted
π Precommit & Commit Stages
To prevent manipulation, miners submit results in two phases:
π Precommit: Submit a cryptographic hash of their result
π Commit: Reveal the final output & validate against Precommit
This ensures transparency & result integrity.
π― Finalizing AI Task Results
Once the task is completed:
β Results are stored on IPFS
β The Router Node retrieves the CID and forwards it to the user
β The session updates dynamically for future requests
π Web2 Users: Real-Time AI Results
For Web2 users, Cortensor provides:
β REST API & WebSocket support for real-time responses
β AI results delivered instantly via standard Web2 services
This ensures fast, seamless AI execution for traditional apps.
π Web3 Users: Decentralized AI Access
For Web3 users, Cortensor provides:
β Smart contract-based result retrieval
β Fully trustless & decentralized execution
β No reliance on centralized APIs
Bridging AI inference into on-chain ecosystems.
π Security & Data Privacy
Cortensor ensures AI task security with:
β Encrypted data flow between nodes
β Session-based ephemeral states
β On-chain transparency via smart contracts
AI execution remains secure, private, and trustless.
β‘ Router Node Scalability
The Router Node is designed for scalability, handling:
β Web2 & Web3 API requests
β AI task scheduling & routing
β Load balancing between miner nodes
This ensures efficient AI inference across different use cases.
π€ Optimized AI Task Execution
With Node Reputation, Ephemeral Nodes, and Router Nodes, Cortensor ensures:
β AI tasks are processed quickly & efficiently
β Users get low-latency AI inference
β The network remains decentralized & scalable
π Learn More: AI Task Execution Flow
Dive deeper into User Interaction & Node Communication in Cortensor by checking out our full technical documentation:
π docs.cortensor.network/technical-architecture/user-serving-and-node-communication
π Like, Share & Join the discussion! π
#AI #DePIN #Cortensor #UserInteraction