OpenGradient x EZKL: Scaling Trustworthy Decentralized Computational Proofs
We want to showcase one of our users, OpenGradient, and how they are using EZKL to scale trustworthy decentralized computational proofs.
OpenGradient is a decentralized AI end-to-end platform for open-source model hosting and application deployment. By developing tooling and a feature-rich platform that makes AI workflow development both secure and seamless, OpenGradient empowers developers to build intelligent and optimized AI-powered applications.
But for you dear reader, what does this mean? Let’s walk through what using a decentralized AI network actually means in practice:
You have an AI model you want to run - say, a price prediction model for cryptocurrency trading
Instead of running this model on a centralized server (like AWS), you’re using a network of independent computers around the world
When you submit your input data (like recent price history), your computation request is distributed to one or more nodes in this network
These nodes process your data through the AI model and return the results
You receive your prediction without needing to maintain your own infrastructure
This might sound great, but it introduces a critical challenge: trust. Here’s why verification matters:
Imagine you’re making million-dollar trading decisions based on these predictions. Without verification, any node in the network could:
Skip running the complex AI computations entirely
Return random numbers instead of actual predictions
Manipulate the results to their advantage
Run a different, simpler model to save on computing costs
This is where EZKL’s cryptographic proofs become essential. When a node returns results, it must also provide mathematical proof that:
It ran exactly the AI model you specified
It used your exact input data
It performed every single computation correctly
The results weren’t tampered with
Given EZKL’s flexibility in compiling diverse models to SNARKs, it represents an ideal complement to secure OpenGradient’s robust decentralized platform.
What has been built
Certain applications demand absolute cryptographic security provided by EZKL and SNARKs, particularly in decentralized finance:
Automated market makers have long sought dynamic pool fees that adapt to market conditions, rewarding liquidity providers for assuming additional risk. The work here leverages EZKL and OpenGradient to generate a dynamic fee mechanism for Uniswap V3.
Risk management in decentralized finance—such as lending protocols—can utilize volatility estimations to provide more secure and efficient loans, potentially addressing the longstanding capital inefficiency stemming from over-collateralized lending models.