From developing AI for Google to creating one of the world’s fastest blockchains: the wild story of NEAR Protocol 🚀
In 2017, engineer Ilya Polosukhin was working at Google when he and a team of researchers published a modest paper titled “Attention Is All You Need.” At the time, no one suspected that this work would change the world. In it, they described the Transformer architecture for the first time—the very technology that powers ChatGPT, Claude, and Midjourney today. That’s right: NEAR co-founder Ilya Polosukhin literally helped create modern AI.
But the story of NEAR itself began when Ilya left Google and launched the NEAR.AI startup with former Microsoft engineer Alexander Skidanov.
The plan was ambitious: teach artificial intelligence to write software code on its own. To do that, they hired top programmers from around the world. But when it came time to pay their salaries, the founders ran into a harsh reality.
Existing blockchains like Ethereum were too slow, too expensive, and wildly complicated for ordinary people. To send a payment, you had to deal with incomprehensible addresses made up of a jumble of characters. Realizing that no ideal payment system existed for their AI project, the developers simply snapped. They put AI aside and decided to write their own blockchain from scratch.
The engineers approached the project with a sense of humor, so they gave the network’s core technologies fantasy-RPG-style names:
• The Doomslug algorithm (literally, “Hell Slug”) handles instant transaction approvals.
• And Nightshade (“nightshade,” a poisonous plant) handles the network’s parallel division into parts (sharding).
The result was NEAR Protocol: a network that replaces complex hash addresses with simple names (for example, username.near), processes transactions in 1 second, and costs 10,000 times less than Ethereum.
The irony is that the project, born because an AI startup couldn’t find a convenient blockchain, is now returning to its roots. NEAR is currently actively developing its AI blockchain initiative and building its own giant artificial intelligence model with 1.4 trillion parameters. Full circle.
$NEAR Tap to trade here 👇
