Starknet introduces AI agents that can perform on-chain activities

Ethereum layer-2 blockchain Starknet will soon allow artificial intelligence (AI) agents on the blockchain that can autonomously perform on-chain activities for users, such as profit optimization and portfolio reallocation.

Giza, a company seeking to bring AI to blockchain, raised $3 million in a pre-seed round led by CoinFund in July 2023 and is leading the effort. Co-founder Renç Korzay told Cointelegraph that AI agents will be deployed on Starknet by the end of June, with other chains to follow.

Korzay said agents can take on four activities: The first is the return rate aggregation. In other words, “find where the highest returns are and deposit in various return paths.”

In a later release, the AI ​​agent will “become a robo-advisor and figure out the best portfolio allocation, taking into account the user’s risk appetite,” Korzay said.

During the beta phase, the Giza team will audit and approve agent strategies to ensure security. In the future, agents will be completely permissionless through the Giza protocol, which provides both cryptography and cryptocurrency economic security to ensure exchanges with minimal trust between developers and users.

They added that they can autonomously provide liquidity to decentralized exchanges or make leveraged price bets and “anticipate price movements” to adjust their cryptocurrency holdings “according to the strategy set by the user.”

Yield-optimized AI agents will be deployed first, with other types to follow, according to a May 9 press release shared with Cointelegraph.

Billed as “low-cost and low-calculation,” it allows you to buy and sell assets to maximize your returns, allowing you to use them directly for transactions and maintain equal weighting in the portfolio of lending platforms that can underwrite them.

“Every account on Starknet is a smart contract through the network’s native account abstraction feature, which allows proof verification directly within the contract,” Giza explained.

The AI ​​model is built in Cairo, the Starknet programming language, and uses zero-knowledge scalable transparent knowledge arguments and zero-knowledge proofs (ZK proofs) created by Starknet development company StarkWare to “verify complex computations.”

ZK Proof allows information to be shared and authenticated while maintaining user privacy.

Related: Philosopher Yuval Noah Harari warns of the risks AI poses to finance.

Giza co-founder Cem Dagdelen said in a statement that the company expects agents to “become a new application layer for Web3 that abstracts smart contract risk and complexity from users.”

However, agents are not limited to cryptocurrencies and blockchains. According to Giza, using the model to ensure “the validity of the results” could also help “highly regulated sectors” reduce supervision costs.

Ava Labs founder Emin Gün Sirer said at a recent blockchain conference that he believes AI-based smart contract coding is five to 10 years away, and that its arrival could bring “billions of new (blockchain) users on board.”

Meanwhile, Ethereum co-founder Vitalik Buterin warned developers that adding AI to blockchains “is worth handling with caution,” noting that AI could help humans understand what’s happening on the chain. They can, but they caution that they should not be tasked with enforcing smart contract rules.

magazine: How to Control AI and Incentivize Humans with Cryptocurrency