Trust Wallet has introduced Agent Kit, a new developer-focused product designed to add an AI transaction layer to self-custody wallet infrastructure. The launch allows developers to build AI agents capable of carrying out real cryptocurrency transactions across more than 25 blockchains, while preserving a core principle of crypto ownership: users remain in control of their wallets and assets.
The release marks a significant step beyond passive AI tooling. Rather than limiting agents to reading wallet data or monitoring prices, Agent Kit enables them to take action onchain under rules defined by the user. In practical terms, that means AI can now be integrated into trading and wallet workflows in ways that are automated, but still bounded by explicit permissions.
Two operating models for AI agents
Trust Wallet said Agent Kit supports two main setup options. In the first, developers can create a dedicated wallet specifically for AI agent activity. Users preconfigure permissions inside that wallet, allowing the agent to execute recurring or rules-based strategies without asking for approval every single time. The company highlighted use cases such as dollar-cost averaging, limit orders, and price alerts.
The second model is designed for users who want tighter oversight. In this configuration, the AI agent connects to an existing Trust Wallet through WalletConnect, proposes a transaction, and waits for the user to approve it before anything is executed. That approach gives users a way to experiment with AI-assisted execution while maintaining manual review over each action.
Trust Wallet’s message is clear across both setups: the wallet remains self-custodial, and the user does not give up ownership of funds in exchange for automation. That distinction is important at a time when many AI finance products are trying to balance convenience with user safeguards.
From read-only data access to onchain execution
The new toolkit builds on the Trust Wallet developer portal launched the week before. That portal initially provided read-only access to crypto data across more than 100 blockchains, including live prices, token metadata, and onchain risk signals. Agent Kit extends that foundation by moving from observation to execution.
In other words, developers can now connect AI systems not only to market and wallet intelligence, but also to transaction capabilities. This makes the platform more relevant for builders working on autonomous trading tools, smart portfolio assistants, and personalized crypto automation products.
At launch, the supported networks include EVM-compatible chains, Solana, Bitcoin, BNB Chain, Cosmos, TON, Aptos, Tron, NEAR, and Sui. Trust Wallet described the rollout as one of the broadest multi-chain AI wallet infrastructure offerings currently available.
Developer tooling and built-in wallet functions
On the technical side, Agent Kit integrates the Model Context Protocol (MCP), a standard increasingly used by developers to connect AI systems with external tools and services. The toolkit is also accessible through a command-line interface, which should make it easier for developers to test and deploy agent workflows without building a full user-facing application first.
According to Trust Wallet, developers can go from account creation to a running AI agent in as little as 15 minutes. That onboarding claim suggests the company is aiming to reduce friction for teams experimenting with AI-powered crypto products.
The out-of-the-box feature set includes token swaps, limit orders, automated strategies, ENS resolution, ERC-20 approvals, message signing, portfolio tracking, wallet auto-locking, and a REST API for deeper integrations. Collectively, those functions give developers a fairly broad set of primitives to build assistants, trading bots, and workflow automation tools on top of self-custody wallets.
Trust layer for AI in finance
In a statement, Trust Wallet CEO Felix Fan said AI agents need a trusted layer before they can safely handle financial tasks on behalf of users. Agent Kit, he said, is meant to provide developers with the tools required to build agents that can operate on real wallets while staying inside rules defined by the user.
That positioning reflects a larger debate now taking shape in crypto and AI circles. The appeal of autonomous agents is obvious: they can monitor markets continuously, react to predefined conditions, and reduce the manual burden on users. But the risks are equally clear if the system executing those actions is opaque, over-privileged, or reliant on custodial control. Trust Wallet is trying to answer that challenge by framing Agent Kit as infrastructure for permissioned autonomy rather than unrestricted automation.
Part of a broader AI-agent push in crypto
The launch comes amid a wider industry shift toward services tailored for the emerging agent economy. Interest in AI agents has accelerated sharply, and companies across the crypto sector have begun releasing tools and infrastructure aimed at this segment. The article notes that firms such as Circle, Binance, and Coinbase have all introduced products or initiatives focused on the space.
In that context, Trust Wallet’s differentiation lies in the self-custody angle. Many crypto users are willing to explore AI-assisted workflows, but not at the cost of surrendering wallet ownership. By giving developers a way to build transaction-capable agents around explicit approvals and permission boundaries, Trust Wallet is betting that self-custody can remain central even as AI becomes more active in execution.
Long-term roadmap: wallet-native AI and an agent marketplace
Trust Wallet said its app has now been downloaded more than 220 million times, underscoring the scale of the installed base it could eventually bring AI functionality to. The company described its broader ambition as becoming self-custody infrastructure for AI-powered finance—a base layer where artificial intelligence can participate in crypto workflows without forcing users to relinquish ownership of their assets.
Over the coming months, Trust Wallet plans to bring AI features directly into the consumer app itself. The roadmap includes in-wallet analysis, automated strategies, and personalized alerts. These additions suggest the company is not limiting AI to a developer product but intends to surface the technology directly to end users.
Trust Wallet also plans to launch a standalone Agent Marketplace, where developers will be able to publish reusable agent strategies and trading bots. Users would then be able to deploy those agents from within the wallet. If executed successfully, such a marketplace could become a distribution layer for AI-native crypto tools, although its effectiveness will likely depend on transparency, review standards, and clear permission management.
What the launch signals for the market
The debut of Agent Kit highlights a notable change in crypto product design. The industry is moving beyond dashboards and alerts toward systems that can actually act on behalf of users. The challenge is to introduce that capability without eroding the principles of self-custody, consent, and verifiable control.
Trust Wallet’s new toolkit does not eliminate the trade-offs around AI-assisted finance, but it does offer a clearer framework for how those systems might operate within user-defined boundaries. For developers, it opens a path to building transaction-enabled AI products across a wide multi-chain environment. For users, it points to a future where wallet software may become less of a passive storage tool and more of an intelligent operating layer for onchain activity.
As the race to build agent infrastructure continues, the companies that gain traction may be the ones that can combine automation with trust. Trust Wallet’s latest launch is an early but important attempt to define what that balance could look like in a self-custodial ecosystem.