Dubai, UAE, Token 2049: Industry leaders Ben Zhou (Bybit), Jansen Teng (Virtuals Protocol), Lia Muller Pena (BitGPT), and Alex Svanevik (Nasen) explored the rapidly evolving world of AI-driven wallets and autonomous agents in crypto. The conversation opened with the idea that AI wallets are quickly becoming more than just digital storage-they’re evolving into autonomous agents capable of executing transactions, managing assets, and even making trading decisions on behalf of users. This shift was likened to how, upon turning 18, people are given bank accounts or credit cards; now, digital agents are being delegated similar powers, but in a purely virtual, programmable environment.
A major focus was on how AI integration can reduce friction for users. Tasks that were once cumbersome-like remembering seed phrases or navigating multiple trading steps-are now being streamlined by AI, making the crypto experience more accessible and less intimidating for newcomers. However, this increased autonomy also raises important questions about trust and control. While some solutions require explicit user approval for each transaction, others envision a future where agents can act independently within set boundaries. The panelists agreed that finding the right balance between automation and human oversight is crucial, especially when significant amounts of money are at stake.

The discussion then turned to where these AI agents are most active. Currently, decentralized exchanges (DEXs) are the primary playground for AI agents, thanks to open APIs and abundant liquidity. While institutional trading firms have long used AI on centralized exchanges, retail users are still hesitant to hand over full control to autonomous agents. Trust, transparency, and the ability to clearly express user intent remain essential hurdles to overcome before mass adoption can occur.
A recurring question was whether AI agents are fundamentally different from traditional trading bots. While both automate trading, the panelists noted that true AI agents are more adaptive, capable of learning from real-time data and sentiment, rather than simply following pre-set rules. Still, for many current applications-like arbitrage or rebalancing-the distinction can be blurry, and sometimes the term “AI” is used more as a marketing buzzword than a technical differentiator.
Safety and liability were also hot topics. What happens if an AI agent makes a costly mistake? The panelists discussed technical safeguards, such as using smaller, fine-tuned language models with strict boundaries, and the potential for insurance products that cover agent-driven activity. There was even talk of multi-agent systems, where one agent proposes a transaction and another challenges or verifies it, adding extra layers of protection.
As for when AI agents will outnumber humans, opinions varied. Some panelists argued that, for certain repetitive tasks, the number of automated trading agents already exceeds the number of human traders. However, for more complex, intent-driven activity, it may take longer for AI agents to truly outpace humans. Still, the consensus was that as agents become more capable and trusted, their numbers will multiply rapidly. In the near future, it’s conceivable that every human will have multiple agents acting on their behalf, and businesses will deploy fleets of agents for everything from commerce to supply chain management.
The panel also touched on the idea of tokenizing the collective output of groups of specialized agents, creating new forms of collaborative, automated business models. This could lead to “DAOs of agents,” where the value created by autonomous agents is transparently tracked and distributed.
On the product side, companies like Bybit are already integrating AI-driven features, such as TradeGPT, to help users make trading decisions and automate customer support. AI is also being used internally for everything from development to legal translation. There’s a strong push toward privacy-preserving, decentralized AI solutions, giving users more control over their data and reducing reliance on centralized providers.
In closing, the panelists agreed that the path to mass adoption of AI agents hinges on building trust, ensuring transparency, and maintaining user control. While the timeline for AI agents to outnumber humans varies depending on the application, the direction is clear: autonomous agents are poised to become a dominant force in crypto, fundamentally transforming how value is managed, transferred, and created. The panel left the audience with a sense of excitement and inevitability about a future where AI-driven agents are not just commonplace, but central to the digital economy.
For more information, please visit the following:
Website: https://www.josephraczynski.com/
Blog: https://JTConsultingMedia.com/
Podcast: https://techsnippetstoday.buzzsprout.com
LinkedIn: https://www.linkedin.com/in/joerazz/
Leave a Reply
You must be logged in to post a comment.