Українська
Русский
English
Trust Wallet launches AI agents for crypto trading across 25+ blockchains
Trust Wallet, one of the largest non-custodial wallet providers with 220 million users, will allow AI agents to independently execute swaps and trading operations across more than 25 blockchain networks.
“The agent economy will arrive sooner than anyone expects,” said Trust Wallet’s new CEO Felix Fan, formerly Chief Product Officer at OKX. “Developers can now build agents that operate real wallets within user-defined rules.”
Two modes — two levels of trust
The system operates in two modes.
The first mode allows users to create a separate wallet for AI activity with predefined limits: the agent independently performs routine tasks — recurring purchases, signal tracking, limit order execution — without requiring approval for each action.
The second mode, described as unique by Trust Wallet, connects the agent to an existing wallet: the agent proposes a transaction and waits for user approval.
The built-in functionality includes swaps, automation, limit orders, portfolio tracking, and risk assessment. Supported networks include Ethereum-compatible chains, as well as Solana, Bitcoin, TON, Tron, NEAR, and Sui.
From an Ethereum wallet to 220 million users
Trust Wallet was founded in 2017 in Bahrain as a mobile wallet focused on Ethereum. The following year, Binance founder Changpeng Zhao acquired a controlling stake.
Over the past three years, the user base выросла from 40 million to 220 million, while annual losses of around $10 million have nearly been eliminated as of 2025.
Last week, the company launched a developer portal, granting AI systems read-only access to data from over 100 blockchains — including prices, token data, and risk signals.
Wallet as an intelligent assistant
Trust Wallet’s strategy is to transform crypto trading into an intent-based experience rather than manual interaction.
Instead of forcing users to navigate swap interfaces, staking dashboards, and order settings, the platform simplifies the process: users express what they want — for example, staking Ethereum or converting $100 into Bitcoin — and the system handles execution.
“We abstract complexity and deliver a result-oriented experience,” Fan added.
In the coming months, AI features will be integrated directly into the wallet: analytics, automated strategies, personalized notifications, and transaction suggestions with optional manual confirmation.
A marketplace for agents is also planned, where developers can publish ready-made strategies and trading bots for in-app use.
At the same time, Fan acknowledges that risks remain. Early adopters will likely be those willing to experiment and tolerate uncertainty.
“We are simply helping users do what they may already be doing manually. But we will make it clear: there is a risk that an agent may use funds in ways the user did not intend.”
AI opinion
The dual-mode architecture of Trust Wallet — a segregated wallet with predefined limits and an agent proposing transactions for approval — effectively mirrors traditional capital management: discretionary vs advisory mandates.
The key issue is not interface convenience but how the agent behaves in exceptional scenarios beyond predefined rules.
Practice shows this is where the main risk lies: in one case, a trading agent misinterpreted a number format and sent $441,000 to an unintended recipient in a single transaction.
Trust Wallet is not alone in this space — Coinbase has already launched similar infrastructure. Competition for developers will likely evolve into a race for functionality, raising the question of whether innovation will outpace the development of safety standards for agent-based systems.
