AI Agents Are Transforming DeFi as User-Control Models Gain Attention

Key takeaways

• The AI-powered crypto sector has expanded significantly, with total value locked rising over 340% and more than 550 projects now active globally.
• DeFi automation is shifting toward fully autonomous agents, though concerns around oversight and risk management remain unresolved for many users.
• New platforms are introducing user-approval frameworks that aim to combine AI efficiency with greater control over on-chain transactions.

AI-driven automation accelerates across DeFi ecosystems

Artificial intelligence is becoming a central component of decentralized finance, moving beyond analytics into direct execution of on-chain activities. Recent data shows that the AI-focused crypto segment has grown by more than 340% in total value locked, with over 550 projects collectively reaching a market capitalization of $4.34 billion.

This growth is unfolding alongside a broader DeFi market valued at approximately $89 billion. Within this environment, AI agents are increasingly used to automate tasks such as managing positions, reducing liquidation risk, and executing cross-chain transactions without manual input.

Several technical developments have supported this shift. Advances in natural language processing allow users to interact with blockchain systems more intuitively, while improvements in infrastructure have reduced transaction costs and increased throughput. For example, Ethereum layer-2 transaction fees have dropped to under one cent, and network capacity has expanded significantly over the past five years.

However, as automation expands, concerns around control and risk exposure are becoming more prominent. Many existing AI-driven tools operate on an “autopilot” model, where users set parameters and agents execute trades independently. This approach has gained traction, with adoption of AI-based portfolio tools increasing by around 300% since 2025.

The limitations of this model became evident during periods of market stress. In October 2025, extreme volatility triggered approximately $1.7 billion in liquidations across Ethereum and related networks. In cases where portfolios were managed by fully autonomous agents, the lack of adaptive oversight meant positions continued to follow preset rules despite rapidly changing conditions.

New approaches emphasize user oversight in automated systems

As the sector evolves, some platforms are exploring alternatives to fully autonomous execution. One emerging approach focuses on maintaining user involvement in every transaction, even when AI handles the underlying logic.

CoinFello is one example of this model. Instead of executing actions automatically, the platform presents each transaction for user approval before it is finalized. The system interprets natural language instructions—such as requests to rebalance assets or manage loan risk—and translates them into structured on-chain operations, which users can review before confirming.

This design aims to address a key concern among DeFi participants: maintaining control over capital while still benefiting from automation. The platform also connects to EVM-compatible wallets and supports account creation through standard methods like email or phone, while keeping assets under user custody at all times.

The distinction between automated and user-approved systems is becoming more relevant as competition intensifies. With hundreds of AI agent projects entering the market, platforms that can balance usability, automation, and control may be better positioned to retain users beyond initial adoption phases.

At the same time, the broader trajectory of AI in DeFi continues to accelerate. Agent capabilities have improved rapidly, enabling more advanced portfolio management and real-time analysis than was possible just a year and a half ago. Projections suggest that AI-driven infrastructure in this space could exceed $52 billion by 2030.

The bottom line

AI is reshaping how decentralized finance operates, introducing new levels of automation and efficiency. At the same time, the debate over control versus autonomy is becoming central to the sector’s development. As platforms experiment with different models, solutions that integrate user oversight with AI execution may play a key role in defining the next phase of DeFi.