Near wants AI to code your next crypto app

What if you could build a complete crypto startup without writing a single line of Solidity?

Near co-founder Illia Polosukhin revealed in an interview with Cointelegraph that the company is developing an AI agent designed to code decentralized applications (DApps) end-to-end.

The goal is simple: let non-technical visionaries describe what they want in plain English, and have the AI handle the rest.

Prompting a DApp into existence

According to Polosukhin, Web3 is currently trapped in a bottleneck where developer talent is scarce and consumer-facing applications are rare.

Near wants its upcoming AI tool to turn anyone with a viable concept into a project founder.

Polosukhin walked through an example of building a freelance marketplace. The AI agent would draft user stories, set up freelancer profiles, write the smart contracts, and construct the front-end user interface.

It would also generate the middleware that stitches the front end to the blockchain. Founders can step in at any point to tweak code or prompt the agent to bolt on missing features.

The push for user-owned AI

Beyond automated coding, Polosukhin outlined his broader philosophy around artificial intelligence risks.

He argues that the biggest danger right now is centralization. When massive tech corporations control frontier models, their fiduciary duty forces them to optimize for engagement and revenue.

Because language models excel at persuasion, centralized systems naturally manipulate people into spending more time on their platforms. To Polosukhin, this is not an evil conspiracy, but an inevitable corporate incentive problem.

His proposed counterweight is user-owned AI. These would be open-source models designed explicitly to protect personal data, safeguard assets, and optimize for the user's economic success.

Polosukhin believes Web3 architecture is uniquely suited to govern these models, putting Near in prime position to build the underlying rails.

Why it matters

For non-technical creators, the promise of end-to-end DApp generation is massive. Lowering the barrier to smart contract deployment could spark a wave of micro-applications and cut down development costs dramatically.

However, we think caution is warranted. Generating front-end layouts is relatively low risk, but smart contracts handle real user funds. Even a minor hallucination or subtle logic flaw in generated contract code can result in catastrophic financial exploits.

Unless Near pairs this generative agent with bulletproof automated auditing, users may hesitate to deploy mission-critical contracts built entirely by AI.

Strategically, this is a smart pivot for Near. Competing strictly on transaction speeds has hit diminishing returns across layer-1 blockchains, and leaning heavily into AI gives the network a fresh narrative to attract developers.

Polosukhin's critique of big-tech AI incentives is spot on, but running performant, user-owned AI on decentralized rails remains an enormous engineering hurdle. It looks like an ambitious roadmap, but whether it becomes everyday infrastructure depends entirely on the agent's real-world execution.