AI Agents Power New Programmable DeFi: Web3's Next Evolution
📌 The Rise of AI Agents in Web3
💱 We're witnessing a confluence of technologies reshaping the Web3 landscape. Just as mobile, social, and cloud technologies defined the last era, AI, crypto, and software agents are poised to usher in a new age of decentralized finance (DeFi). This evolution emphasizes user intent, where autonomous agents act on behalf of users to navigate the complexities of DeFi.
According to DappRadar, AI's integration into crypto has rapidly transitioned from a novelty to a fundamental component. Large language models (LLMs) are now used to summarize governance proposals, agents are rebalancing portfolios, and bots are executing on-chain strategies in real-time. By June 26, 2025, AI-agent projects had already raised $1.39B year-to-date, significantly surpassing the funding rate of 2024. This surge in investment underscores the growing confidence in AI's potential within the crypto space.
📌 The Symbiotic Relationship Between AI and Blockchain
🔗 Chris Dixon of a16z crypto highlights the complementary nature of AI and blockchain. Blockchain provides essential elements like ownership, credible commitments, and identity, which are lacking in primitive AI systems.
As Dixon notes, “AI needs blockchain-enabled computing.” This synergy is crucial for creating open markets for compute, data, and content. NVIDIA’s Jensen Huang echoes this sentiment, viewing AI as the catalyst for “a new industrial revolution,” which will necessitate new user layers and automation patterns in finance.
📌 From Applications to Autonomous Agents
📝 The future of DeFi is moving towards a model where users simply state their intent, and an autonomous agent takes over, composing the necessary stack—data, liquidity, risk checks, and settlement—to execute the desired outcome. This concept, explored in research on agentic systems and the “Agentic Web,” envisions a world where agents autonomously pay for data and services, coordinate via smart contracts, and transact without constant human intervention.
💱 Developer tooling is evolving to support this shift. Frameworks like elizaOS demonstrate how to connect LLM agents to wallets and DeFi actions, enabling natural language commands such as “transfer” and “swap.” This suggests a future where the application itself becomes an agent orchestrator.
📌 The Data Fragmentation Challenge
💱 AI agents rely on reliable, low-latency data to function effectively. However, Web3 currently suffers from data fragmentation across different chains, schemas, and sources.
Raw chain data is time-ordered and scattered, requiring specialized indexing, subgraphs, replication, and ETL pipelines to extract meaningful insights. Providers like Goldsky and The Graph are addressing this issue, but even they emphasize the need for cross-chain mirroring, real-time streaming, and composable subgraphs to support complex applications that agents will continuously demand. Independent analyses confirm the significant costs of data fragmentation for DeFi risk management and user experience.
Takeaway: As the user interface transforms into an intent box, the primary challenge shifts to building a programmable data layer that normalizes on-chain and off-chain context, provides deterministic APIs for agents, and supports low-latency computation for alerts, scoring, and routing across chains.
📌 Why AI Agents Are a Natural Fit for DeFi
💱 DeFi's machine-native environment—characterized by transparent ledgers, programmable liquidity, and composable contracts—makes it an ideal testing ground for autonomous agents. These agents can perform various critical functions:
- Trade and rebalance: Agents can execute structured prompts like “sell long-tail assets into ETH if volatility exceeds X.”
- Scan risks: They can continuously monitor contract anomalies and oracle drift, factoring these risks into execution strategies.
- Arbitrage and MM: Agents can perform arbitrage and market-making across decentralized and centralized exchanges without user interface friction.
- Govern: They can draft proposals and simulate outcomes using on-chain and forum data.
💱 Academic research on autonomous AI agents in DeFi supports these roles, linking agent decision-making to market microstructure and governance design. Vitalik Buterin suggests that AI's most viable role is as a “player” in crypto games, which seamlessly translates to market activities.
📌 The Emerging Landscape of Chat-Based DeFi Platforms
💱 Several chat-based or agent-first products are emerging, each offering a unique approach to intent-centric execution:
- HeyElsa: An AI crypto co-pilot that uses natural language/voice to route, bridge, swap, and lend across chains with MPC-secured wallets and safety rails. Its USP is a unified chat/voice control combined with a custody model designed for mainstream user experience.
- Kuvi.ai: Self-described as Agentic Finance, enabling text-to-trade execution across DeFi. It positions agents as solvers that connect user intent to settlement. Its USP is an end-to-end intent pipeline with cross-domain ambitions in finance, identity, and gaming.
- Igris.bot: Focuses on destination-based swaps, allowing users to specify their desired outcome (e.g., "end with 2 ETH on Base"). The system determines the portfolio source, route, and fees between chains. Its USP centers on destination rather than source, reducing user decision load and tapping latent portfolio liquidity.
- Defi App: Offers explicit intent-based swaps via solver/relayers, routing across multiple aggregators/DEXs. Its USP is native intent-based execution using a solver model, where off-chain solvers/relayers compete to route across various liquidity sources.
- AskGina.ai: An AI wallet companion that analyzes holdings and executes on-chain transactions from chat. Its USP is an AI wallet companion with a chat interface that understands your portfolio and provides tailored insights.
📌 The Infrastructure Required for the Agentic User Layer
⚖️ As agents become the new user interface, infrastructure must adapt to machine requirements:
- Programmable Data Layer: This involves cross-chain ingestion, normalized schemas, real-time replication/mirroring, and deterministic APIs that agents can consume.
- Latency-aware Compute: Triggers for price/volatility/MEV risk, agent policy evaluation, and pre-trade checks are crucial.
- Identity & Permissions: Wallet-bound permissions, cryptographic attestations ("proof of personhood/humanity"), and policy guards around agent autonomy are essential, leveraging blockchain’s strengths as Dixon highlights.
- Safety Rails: Vitalik Buterin’s cautions regarding restricted APIs, circuit breakers ("kill switches"), and alignment layers must be prioritized.
📌 Why This Is Important Now
🔗 The intent-centric model is gaining momentum, as users increasingly expect to simply state their goals and have agents handle the underlying processes. The current paradigm—clicking across bridges, DEXs, and dashboards—is unsustainable for the next 100M users. The architectural solution involves creating open rails for ownership and programmable data, allowing numerous agents to compete on user value, rather than relying on a few closed super-apps.
When major waves converge, they “complement each other and work together.” AI offers creativity and automation, crypto provides open ownership and incentives, and new devices (from phones to wallets) drive distribution. Together, they form a user stack that defaults to agents.
📌 🔑 Key Takeaways
- AI agents are rapidly transforming DeFi by automating tasks, optimizing strategies, and simplifying user interactions.
- Data fragmentation remains a key challenge, requiring robust programmable data layers to support AI agent functionality.
- The convergence of AI, crypto, and agents signifies a shift towards intent-centric models, improving scalability and user experience in DeFi.
- Emerging chat-based platforms demonstrate the potential of AI agents to streamline DeFi operations, offering users intuitive interfaces and automated execution.
- Infrastructure must evolve to meet the demands of AI agents, including programmable data layers, latency-aware compute, and robust security measures.
🔮 Thoughts & Predictions
The rapid integration of AI agents into DeFi is more than just a trend; it's a paradigm shift. The challenge lies in overcoming data fragmentation to unlock the full potential of these agents. Expect to see significant investments in cross-chain data solutions and AI-powered risk management tools over the next 12-18 months. These developments will likely lead to increased institutional adoption as DeFi becomes more accessible and secure, potentially pushing the total value locked (TVL) in DeFi protocols past the $500 billion mark by the end of 2026. The key will be identifying and investing in platforms that prioritize both user intent and robust data infrastructure.
🎯 Investor Action Tips
- Identify and research DeFi projects actively integrating AI agents and prioritizing data accessibility and interoperability.
- Monitor the development and adoption of cross-chain data solutions like Goldsky and The Graph, as these will be critical for AI agent functionality.
- Evaluate the safety and security measures implemented by platforms offering AI-powered DeFi tools, particularly regarding permissions and safety rails.
- Explore chat-based DeFi platforms and assess their potential for simplifying user interactions and automating complex tasks.
🧭 Context of the Day
Today's focus is on the escalating race to build robust, AI-driven DeFi platforms, highlighting the critical need for enhanced cross-chain data solutions.
