When AI meets Web3: A Singapore firm chasing full automation in crypto trading

Nansen isn’t a name that makes most people think of AI – even if its URL is Nansen.ai.

Founded by CEO Alex Svanevik in 2020, Nansen is known in Web3 for its ability to discover the hidden identities behind crypto wallets, which helped make the Singapore-based firm one of the top analytics platforms in the blockchain industry.

Alex Svanevik, CEO and co-founder of Nansen.ai, a firm that uses machine learning to identify holders of Web3 wallets. / Photo credit: Alex Svanevik, enhanced with the use of Photoshop’s Generative Fill.

But Nansen has made no secret about its strategy for discovering the “who’s who” on-chain. Machine learning has always been a part of the process, with Svanevik calling the firm’s use of AI to label Web3 wallets “the first use case.”

“It’s actually something we’ve been doing for years,” co-founder and CEO Alex Svanevik tells Tech in Asia. “We’ve used AI to label addresses, to price NFTs, and to classify spam tokens. But now we’re ramping up our AI efforts across the board.”

Nansen’s focus on AI went into overdrive in the first quarter of 2023 when Svanevik set a staff goal of dedicating 5,000 hours to using the tech, with the goal of driving adoption and innovation, he says. The CEO increased the goal to 6,000 hours the next quarter.

One of the results of the firm’s dive into generative AI is Signals, which was launched in January this year. Using machine learning, it can analyze transactions on between 10 and 50 blockchains, depending on what users are looking for.

The idea behind this service, Svanevik explains, is to help clients filter out the millions of irrelevant transactions – Ethereum alone currently sees about 1.2 million transactions a day – to find the 10 or so that are actually relevant. This is based on the client’s investment strategy, its portfolios, and the “general newsworthiness” of what’s happening online.

Signals will then notify users when it finds a transaction the client might be interested in.

See also: The good, the bad, and the hallucinations: What SEA startups think of AI

“It’s a bit analogous to machine learning in the context of Google or Amazon or even ads that fit with Facebook,” Svanevik says. “You’re trying to put something in front of people that they will click.”

Eventually, the goal of Signals will be to not only identify possible trades to Nansen’s clients, but to suggest them and possibly execute them.

No one digging around in the data

Self-executing trades, however, isn’t what made Nansen develop Signals.

Svanevil says that based on feedback, users didn’t want to spend time digging around in Nansen’s data. They “wanted something that was more easily digestible,” which led the Web3 firm to build Signals to serve on-chain events to users “on a silver platter.”

AI to bring it back in the black?

Need humans in the AI loop

Can AI do crypto trading for you?