Spectral Labs, a cryptocurrency development company focused on artificial intelligence, wants to make it easy for non-programmers to build on blockchain. On Tuesday, the company will release Syntax, an AI app that anyone, programmers and non-programmers alike, can spin up. smart contract Compatible with Ethereum and dozens of other blockchains.
While Syntax represents an impressive technical achievement, Spectral still faces the hurdle of convincing users to entrust their valuable digital assets to AI.
Chat Those familiar with GPT and similar chatbots will be familiar with the Syntax interface, which leverages cryptographically specific large-scale language models (LLMs) and allows you to have SMS-like conversations with users.
In addition to answering blockchain-related queries such as “What is the circulating supply of ETH?”, the web app also responds to prompts from users with production-ready Solidity code (for other blockchains based on Ethereum and the Ethereum Virtual Machine). (programming languages used in). (EVM) standard.
“Syntax users can compile, debug, and deploy AI-generated solidity code,” Spectral Labs said in a statement, adding that NFTs, arbitrage bots, and rollups are among the technology’s possible use cases. He added that there is.
“For example, let’s say you want to encode an ERC-20 token called ‘ABC’ with a supply of 100 million,” Spectral Labs CEO Sishir Varghese told CoinDesk in an interview. “The syntax generates all the code for the user, who can compile, test for vulnerabilities, test for gas efficiency and code optimization, and deploy the app on-chain directly from the interface.”
AI-powered coding assistants are not new. The engineer here has been generating code using ChatGPT for a year. Also, his AI-powered coding guru at Microsoft, GitHub Copilot, has become a mainstay in the modern programming arsenal since its release in 2021.
“I don’t think there really was an LLM that was fine-tuned for Solidity,” Varghese says.
In addition to being tailored specifically for Solidity development, Syntax is also novel in that it can connect directly to EVM-compatible blockchains. This means that with the push of a button, a user can instantly deploy an “agent” created with Syntax to a blockchain such as Ethereum, Arbitrum, or Coinbase’s Base.
The app can be configured with private keys and other data, allowing agents to control real cryptocurrency funds. This means that agents will be able to buy, sell, and trade tokens as if they were in the program, depending on the content of the program. He was a real human being.
There is always the thorny issue of ‘hallucinations’ when it comes to using LLMs. AI can make things up, and in the case of coding, it can create correct-looking code that is error-prone or behaves in unexpected ways. When AI-generated code begins to handle real money, hallucinations become an obvious problem.
“We tried to introduce some safeguards in terms of vulnerability detection and code efficiency,” Varghese said. “All of these things are kind of already built in. We’re also going to have some pre-built agents that can help detect vulnerabilities as well.”
He added that Syntax was trained on a specially curated set of audited smart contracts, security reports, and developer documentation.
“Creating a high-quality dataset is really important because we want the LLM to be able to reproduce it from a code perspective,” says Varghese. “If you take everything that has ever existed, you are not going to get a good result.”
Spectral Labs’ CEO acknowledged that using an AI tool like Syntax necessarily involves some risk. “I think seasoned developers are still pretty nervous about everything co-pilot related, with the exception of Web2. For Solidity, this will be completely new.” It’s an experience to be honest. ”
Varghese expects retail traders, rather than experienced programmers, to be the primary users of Syntax.
“I don’t think they’ll jump in and trust them,” he said of seasoned developers. “I think what’s even better is that even retailers who have never deployed a contract before or even generated a Solidity code can do this so seamlessly and easily.”
