Alex Shevchenko says he is excited about user-specific generated front-ends and autonomous research agents, which he thinks will drive unprecedented progress in drug development and materials creation.
Shevchenko: Autonomous AI Agents and Blockchain Interoperability to Reshape Tech and Science
While today’s large language models (LLMs) have become “extremely” proficient at language-related tasks, Alex Shevchenko, co-founder and CEO of Aurora Labs, said he is more “excited about the potential of helper agents for software, hardware, and general engineering tasks.” Although the technology is “not quite yet there,” a confident Shevchenko said he is placing a “high bet on user-specific generated front-ends.”
In his written responses to questions from Bitcoin.com News, the Aurora Labs CEO stated that ongoing developments in autonomous research agents position makes him believe that the world on the cusp of unprecedented progress in drug development and the creation of new materials. Furthermore, Shevchenko, a blockchain and high-performance computing expert, said he envisions a future where personal-assistant artificial intelligence (AI) agents perform tasks and achieve complex goals without human supervision.
Meanwhile, the Aurora Labs CEO acknowledges that the early portrayal of blockchain as an innovative technology that would finally disrupt traditional industries may have contributed to the resistance against the technology that persists today. However, he believes that the technology’s transparency—one of blockchain’s core attributes—poses a challenge to organizations and governments that are not accustomed to being scrutinized hence their continuing opposition to the tech.
In the remainder of his responses, Shevchenko also discussed the challenges associated with enabling interoperability between blockchains. Below are all of his answers to the questions sent.
Bitcoin.com News (BCN): One of the reasons blockchain technology initially gained momentum was because it was seen as revolutionary something which left traditional financial institutions feeling threatened. With the benefit of hindsight, do you think that the early positioning of the blockchain technology as disruptive and revolutionary made Web2 firms look at it in a negative light?
Alex Shevchenko (AS): Absolutely. Not only this but also the early applications of the technology – value transfers in dark markets. However, people soon realized that crypto is the worst possible way to transact anonymously since all the info is publicly accessible. And this transparency was also repelling non-financial institutions from the blockchain: the way how many governmental agencies, pharmaceutical companies, and many other businesses used to work – was to keep everything closed behind many doors. Blockchain exposes all of the processes, often making stakeholders in such traditionally closed environments uneasy.
BCN: Enabling blockchain interoperability poses significant security challenges. Interoperability solutions have been vulnerable to hacking, with many cybercriminals exploiting weaknesses in protocols. What do you think contributes to these vulnerabilities, and how can the industry mitigate them?
AS: We are living in a multi-blockchain world and it’s clear this reality isn’t changing anytime soon. As new solutions emerge and others fade away, significant value naturally flows between blockchains. That’s why interoperability solutions usually hold enormous amounts of assets. For example, Rainbow Bridge, launched and successfully operated by Aurora Labs for 4 years without a single hack, was holding more than $1.5B in highly liquid assets (stables, ETH WBTC) on its contracts during the peak of the 2021 bull market. This is not something that college grads are prepared for. And college grads are the medium persona of the industry. Simply put, we’ve grown too much too fast, so security mismanagement is just the growth problem.
Our recipe is to introduce as many security layers as possible, which exponentially reduces the probability of production accidents. These layers are: security best practices education for engineers, automatic testing agents, rigorous code review, pre-merge access to the codebase for auditing companies, multi-party security audits before major releases, beta-testing with integration devs, gradual feature rollout, constant deficiencies monitoring, bug bounties, and in some cases insurance.
BCN: Numerous narratives surround artificial intelligence (AI) agents today. In your opinion, what exciting developments in AI do you think deserve more attention?
AS: LLMs are extremely good at language-related tasks since they are specifically built for this purpose. So the majority of ‘wow’ use cases revolve around text analysis and generation. However, other types of models are emerging, and LLMs themselves are becoming more capable in areas beyond language. I’m very excited about the potential of helper agents for software, hardware, and general engineering tasks. We’re not quite yet there, but I put a high bet on user-specific generated front-ends. Additionally, there’s a lot of development in the area of autonomous research agents – tools capable of creating new materials or drugs and optimizing processes and structures much faster than humans. These powerful advanced tools have the potential to dramatically accelerate scientific progress.
BCN: Do you think AI agents (fully autonomous AI bots) will be able to collaborate to achieve complex goals without human supervision?
AS: Sooner or later, yes. It’s easy for me to imagine a world where a personal assistant AI agent could handle tasks like finding help to open a bank account and make a deal with another specialized AI agent for this piece of work. The human input might be as simple as setting a goal like, “Make my life easier in this new country I just moved to.”
BCN: Aurora is said to be a network of virtual chains on the Near Protocol. Can you explain to our readers what are these virtual chains and how do they open the doors to a multichain world?
AS: Virtual Chains is a concept akin to L2s but with a twist. They can cross-call one another and the mother chain infrastructure. This feature addresses the cold start problem typical of L2s – where a newly launched L2 lacks essential infrastructure. With Virtual Chains, all the infra that you need—like RPCs, indexers, custodians, oracles, bridges, message passing protocols, stablecoins, CEX integrations, DeFi ecosystem, launchpads, NFT marketplaces, explorers, cross-chain primitives, gas abstraction and many-many others—is available right from the first block.
Where infra setup for L2s often takes months (if not years) and costs tens of millions of US dollars of CapEx and millions per year in OpEx; Virtual Chains achieve this at zero cost. This dramatically lowers the barrier for new use cases, enabling projects that couldn’t previously afford L2 or the even bigger L1 costs. I expect lots of great ideas to be generated and quickly tested in 2025—hundreds, if not thousands.
BCN: Aurora has developed the Bitcoin Light Client and Relayer service, enabling interactions between the Bitcoin network and NEAR Protocol. With growing developer activity in the Bitcoin DeFi and Web3 ecosystem, what do Aurora’s efforts to connect Bitcoin and NEAR Protocol mean for Web3 builders and users?
AS: Bitcoin has always been a cornerstone of the blockchain ecosystem and up until now, there were no ways to incorporate this asset in DeFi — simply because Bitcoin script is just too poor in comparison with the EVM capabilities. However, new technologies like Chain Signatures, innovated by the Near Protocol, made it possible to build truly decentralized and permissionless solutions that bridge Bitcoin to other blockchain ecosystems.
With Bitcoin Light Client, we can expect two major types of projects on Near. The first one is focused on utilizing BTC value in DeFi projects (bridging and swapping, using BTC as collateral, etc.) and building next-generation L2-style solutions on top of Bitcoin. And since our technology stack is not only connected to Bitcoin but also to other networks, Aurora and Near become the enablers of using BTC in all other ecosystems.
BCN: Considering how the Internet evolved, moving from Web1 to Web2, how long do you think it would take for the majority of regular Internet users to become familiar with Web3?
AS: I believe that within the next five years, 80% of the world’s population will have some level of exposure (perhaps indirect) to crypto assets. This doesn’t mean that these people will have their seed phrases—a relic of the techie past. But new forms of self-custody (passkeys for example) would result in digital cash adoption. We will forget about the underlying infrastructure, like whether an app runs on Solana or Aurora or Near, the same way we don’t know which cloud provider is used by a particular website. The focus will shift from infrastructure to the products themselves. And communities will form around products, not platforms. The blockchain is an unstoppable revolution.
