Join the Agents Gizathon and build trust-
minimized AI Agents for Web3.
Check out the hackathon summary and winners here.
Go autonomous
with on-chain Agents
Giza Agents is a modular and extensible framework for integrating trust-minimized Machine Learning into on-chain strategy and action. The extensible nature of Giza Agents allows developers to enshrine custom strategies using ML and other algorithms, develop novel agent functionalities, and manage continuous iteration processes.
Giza is building the foundational infrastructure to bring machine learning capabilities to Web3.
Ideal for developers, Giza offers the tools to explore and develop ML solutions for the Blockchain, unlocking the future of decentralized applications.
$40,000
Available in Prizes
$10k Total Prize Pool
Giza is building the foundational infrastructure to bring Machine Learning capabilities on-chain, through integrating ZKML-powered on-chain Agents.
-
Develop an AI Agent using the Giza SDK that manages a smart strategy developed on top of a decentralized protocol. Agents use the Ape framework to interact with smart contracts based on Verifiable ML Model predictions. The Agent must follow a predefined logic, and execute on-chain transactions based on the output of a verifiable ML Model.
Potential Use Cases:
Financial Agents:
Algorithmic structured products
Arbitrage Agents
Insurance Agents
LP Automation AgentsProtocol Agents:
Treasury management Agents
Reputation scoring Agents
Slashing AgentsSocial Agents:
Recommender Agents
Content mod AgentsQualification Requirements:
- Deploy your Agent on-chain, with proof of successful transaction. The Agents can be deployed on any testnet supported by the Giza SDK.
- Agent must be aligned with the predefined objective/problem it is designed to solve.
- Source code hosted on GitHub. Code must be reproducible and easy to execute.
- Extensive documentation (agent business case, functionality, attack vectors, possible improvements)
- Agent must contain algorithmic logic and Machine Learning. The use case must need ZK proofs. Agent must utilize the predictions in a coherent strategy. It must be a feasible problem for an ML model. -
This prize rewards creative and innovative AI Agent concepts, including fully implemented Agents or ZKML models designed for AI Agents that also showcase Agent action design, protocol interaction scope, and specifications for smart contract integration
Qualification Requirements:
- Deployed Agent on-chain, with proof of successful transaction, or deployed endpoint for Verifiable Inference. If you don’t have a fully implemented Agent you must still submit the action design for Agent, Protocol interaction Scope, and Specs for smart contracts
- Source code hosted on GitHub. Code must be reproducible and easy to execute
- Extensive documentation (Agent business case, functionality, attack vectors, possible improvements)
$10k Total Prize Pool
Starknet is a Layer 2 Validity Rollup over Ethereum. It utilizes the power of STARK technology to ensure computational integrity. By validating off-chain transactions, Starknet overcomes scalability limitations, while retaining Ethereum’s security and decentralization.
-
Develop an AI Agent using the Giza SDK that manages a smart strategy developed on top of a decentralized protocol on Starknet Sepolia. Agents use the Ape framework to interact with smart contracts based on Verifiable ML Model predictions. The Agent must follow a predefined logic, and execute on-chain transactions based on the output of a verifiable ML Model.
Potential Use Cases:
- Trading on DEXes
- Portfolio Optimization
- Asset Allocation
- Automated Liquidity ManagementQualification Requirements:
- Deploy your Agent on Starknet testnet with proof of successful transaction.
- Agent must be aligned with the predefined objective/problem it is designed to solve.
- Source code hosted on GitHub. Code must be reproducible and easy to execute.
- Extensive documentation (Agent business case, functionality, attack vectors, possible improvements) -
This prize rewards creative and innovative AI Agent concepts on Starknet, including fully implemented Agents or ZKML models designed for AI Agents that also showcase action design, protocol interaction scope, and specs for smart contracts.
Qualification Requirements:
- Deployed Agent on-chain, with proof of successful transaction, or deployed endpoint for Verifiable Inference. If you don’t have a fully implemented agent, you must submit action design for Agent, Protocol interaction Scope, and Specs for smart contracts
- Source code hosted on GitHub. Code must be reproducible and easy to execute
- Extensive documentation (Agent business case, functionality, attack vectors, possible improvements)
$10k Total Prize Pool
Enzyme Finance is a protocol for smart, tokenized wallets designed to accelerate project outcomes and facilitate on-chain asset management.
-
Develop an advanced on-chain strategy agent that leverages ZKML and the robust features of Enzyme's ecosystem. This agent should automate interactions with Enzyme vaults or other Enzyme products, enhancing functionality and bringing significant automation value to users and managers.
Potential Use Cases:
- Signal-based strategies, especially signals based on ETH and leveraging x/-x accordingly
- Automated copy trading from tradfi data sets including indices, etc.
- Carry trades based on ETH and perps,
- Recursive liquid restaking based on rewards and yield signals.Qualification Requirements:
- Deployed Agent on-chain, with proof of successful transaction, or deployed endpoint for Verifiable Inference. If you don’t have a fully implemented agent you must submit the action design for Agent, Protocol interaction Scope, and Specs for smart contracts.
- Source code hosted on GitHub. Code must be reproducible and easy to execute
- Extensive documentation (agent business case, functionality, attack vectors, possible improvements)
$5k Total Prize Pool
zkSync is a Layer-2 protocol designed to scale Ethereum. With zkSync, it’s trivial for Web3 developers to take advantage of high-speed, low-cost transactions with the same security guarantees as Ethereum.
-
Develop an advanced on-chain strategy agent for Koi Finance on zkSync.
Potential Use Cases:
- Concentrated liquidity pools for Koi
- Liquidity RebalancingQualification Requirements:
- Deploy your Agent on zkSync Sepolia, with proof of successful transaction, or deploy your endpoint for Verifiable Inference. If you don’t have a fully implemented agent you must submit the action design for Agent, Protocol interaction Scope, and Specs for smart contracts.
- Source code hosted on GitHub. Code must be reproducible and easy to execute
- Extensive documentation (agent business case, functionality, attack vectors, possible improvements)
$2k Total Prize Pool
Linea is an EVM-equivalent network, secured with a zero-knowledge rollup to Ethereum, built on lattice-based cryptography, and powered by Consensys.
-
Develop an AI Agent using the Giza SDK that manages a smart strategy developed on top of a decentralized protocol on Linea Sepolia.
Qualification Requirements:
- Deploy your Agent on Linea Sepolia, with proof of successful transaction, or deploy your endpoint for Verifiable Inference. If you don’t have a fully implemented agent you must submit the action design for Agent, Protocol interaction Scope, and Specs for smart contracts.
- Source code hosted on GitHub. Code must be reproducible and easy to execute
- Extensive documentation (agent business case, functionality, attack vectors, possible improvements)
$3k Total Prize Pool
Lambda is data platform for web3 developers. Growing interest to staking and re-staking products makes Lambda be focused on slashing risk data as core feature for automated staking strategies.
-
Qualification Requirements:
- Deploy your Agent on Sepolia, with proof of successful transaction, or deploy your endpoint for Verifiable Inference. If you don’t have a fully implemented agent you must submit the action design for Agent, Protocol interaction Scope, and Specs for smart contracts.- Source code hosted on GitHub. Code must be reproducible and easy to execute
- Extensive documentation (agent business case, functionality, attack vectors, possible improvements)
Hackathon Schedule
Mon, May 20th5:30 pm BST | Launch Event + Hacking with Giza: Welcome to Gizathon Agents | Sign up here! |
Tue, May 21st4:00 pm BST | Full Stack Giza Workshop | |
Wed, May 22nd4:00 pm BST | Mental Models for ZKML and its Applications | |
Wed, May 22nd6:00 pm BST | Fundamentals of On-chain Agents | |
Thurs, May 23rd4:00 pm BST | Agents Gizathon Partner Panel |
Whole Week | Build, build, build! | Sign up here! |
Whole Week | Finalise your builds! | Sign up here! |
Sun, Jun 9th23:59 Your Timezone | Submission Deadline | |
Beginning of JuneTime TBA! | Pitch practises for finalists! | |
Wed, Jun 19th5:30 pm BST | Finale and Prizegiving |
Final push! | ||
Sun, May 5th23:59 in your time zone | Submission deadline! | |
End-JuneWeek of 17th June | Pitch practices for finalists! | |
Wed, June 19th5:30 pm BST | Finale and Prizegiving | Sign up here! |
Upcoming Events
Previous Events
Speakers
Ege Atesalp
ML Engineer
Giza
Fran Algaba
Co-Founder
Giza
Jakub Warmuz
AI Engineer
Giza
Cem Dagdelen
Co-Founder
Giza
Raphael Doukhan
ML/Blockchain Engineer
Giza
Renç Korzay
Co-Founder
Giza
Luca Mossini
Senior Business Development Consultant
Enzyme
Vassilis Tziokas
Head of Enterprise Strategy & Business Development
Matter Labs
Title Sponsor
Giza is building the foundational infrastructure to
bring Machine Learning capabilities on-chain, through integrating ZKML powered on-chain Agents.