Latest news
Bitget launches dedicated trading accounts for AI agents, GetClaw autonomous trading now live
2026-04-01 07:594249
As AI capabilities evolve from information understanding and task execution towards autonomous collaboration and persistent decision making, a new class of participants is entering trading markets: AI agents that can understand objectives, continuously monitor markets, execute strategies, and close the full trading loop.
Today, Bitget is officially launching GetClaw trading capabilities and becoming the first exchange globally to offer dedicated, standalone trading accounts for AI agents. By provisioning an exclusive sub-account for GetClaw, Bitget is moving far beyond AI-assisted trading—we are giving agents the ability to directly enter markets, manage capital, and execute trades.
From assistant tool to autonomous trader: The era of agent-native trading begins
For most of their history, AI tools in trading have operated strictly as decision support. They can pull prices, run analysis, generate commentary, and suggest strategies—but capital management, order execution, position adjustment, and continuous monitoring have always remained fully manual operations.
With the maturation of large language models, task orchestration systems, and autonomous agent capabilities, markets no longer need another analysis assistant. What is required is an agent that can understand intent, decompose tasks, monitor conditions, execute actions, and return results. AI has evolved from a passive support tool into an autonomous trading participant.
This is exactly what GetClaw was built to deliver. With its dedicated isolated account, GetClaw now has full authority to manage capital and execute complex strategies with a high degree of autonomy and flexibility. For the first time, an AI agent is integrated into the trading system as a market participant, not an afterthought plugin.
This is not just another feature launch—this is a fundamental shift in trading infrastructure: exchanges now serve not only human users, but the rapidly growing population of AI agents.
GetClaw: The implementation for agent-native infrastructure
As Bitget's trading agent product, designed for the agent era, GetClaw is not just another chatbot that answers market questions, nor is it a natural language wrapper around an existing trading interface. It is built from first principles for how agents actually operate: Intent recognition as the entry point, task-driven execution as the core, and a closed loop of persistent operation and outcome feedback.
Under this model, GetClaw can own end-to-end execution of the full workflow within user-authorized boundaries: market observation, trigger conditions, order placement, account status queries, and more. It does not stop at giving recommendations—it can continuously participate in trading operations within explicitly defined rules.
The entire GetClaw system is built natively for agents, with deep optimizations for the extreme speed and high-frequency data processing requirements of AI. This is not a human interface with an assistant tool bolted on. We have rebuilt the underlying infrastructure from the code level up, designed explicitly to facilitate interaction between autonomous agents and global markets—not just humans.
GetClaw autonomous trading: Available now
With this launch, users can initiate trading commands directly through conversation with GetClaw on Telegram, and will be guided through sub-account creation, funding, and basic trading operations. Core capabilities available today:
Dedicated GetClaw trading sub-accounts
When a user specifies a trading objective such as "Trade BTC for me", the system will automatically provision an isolated sub-account in seconds, allowing users to get started immediately.
Supported trading types
GetClaw sub-accounts currently support spot trading, USDT-Margined Futures, order placement, and position closing. This already enables a full end-to-end trading workflow through conversational interaction, covering all critical actions from entry to exit.
Multiple trigger modes
Currently, GetClaw supports two main types of trade triggers: Scheduled tasks and immediate order execution.
Trades can be executed either immediately on request or set up as persistent monitoring tasks that execute automatically when predefined conditions are met. This moves trading beyond manual point-and-click operations to an intent-driven, condition-execution model.
Capital transfers
Users can transfer funds into and out of the GetClaw sub-account either via the standard transfer feature under Profile>Assets>Transfer, or directly through the funding page.
For classic account users, funds can be transferred to spot or USDT-M Futures accounts. For unified trading account users, funds are transferred directly to the unified trading account. Once funded, GetClaw can operate within that account.
Real-time account visibility
Users can query sub-account balances, open orders, and current positions at any time directly within the GetClaw conversation.
Eight real-world use cases: Agents that actually trade
24/7 market monitoring & opportunity detection
GetClaw watches markets, signals, and opportunities around the clock, responding immediately at critical moments even when you are offline.
Technical signal-based automated strategy execution
From MACD, RSI, volume breakouts, and multi-indicator confirmation, GetClaw turns technical signals into executable actions so you never miss an opportunity to hesitation.
Trend following execution
Wait for confirmation, enter with conviction, exit cleanly on weakness. GetClaw executes disciplined trend following consistently, every time.
Range-bound & grid execution
No chasing moves in sideways markets. GetClaw automatically executes buy low and sell high, grid strategies, and scalping operations to extract value from quiet, ranging conditions.
News & event awareness & response
From FOMC decisions, CPI prints, breaking news, and narrative shifts—GetClaw continuously monitors for events and positions ahead of market reaction.
On-chain signal & flow monitoring
Detect whale movements, capital flow shifts, and on-chain activity spikes, identifying opportunities before they are fully priced in.
Take-profit/stop-loss & risk management
It is not just about finding opportunities. GetClaw automatically manages TP/SL, drawdown limits, and risk constraints to keep trading disciplined.
Dynamic capital allocation
Not all capital should be deployed at all times. GetClaw dynamically allocates funds based on market conditions, strategy status, and risk profile to optimize capital efficiency.
Security principles: Isolation, least privilege, bounded risk
Security was the foundational design constraint for GetClaw's trading capabilities. Rather than granting agent access to main user accounts, Bitget implemented a dedicated agent sub-account architecture, designed to unlock execution capability while containing, limiting, and clearly defining risk boundaries.
All agent trading activity operates within a completely separate sub-account, with hard boundaries between this account and the user's main balance. All capital usage, positions, and trading activity are contained entirely within this isolated environment, eliminating any possibility of unintended impact on the user's primary assets. This also provides full transparency for auditing, behavior analysis, and risk tracking.
GetClaw does not have unrestricted account access. It operates only with explicit user authorization, within predefined product boundaries, and only inside the dedicated sub-account. Currently, GetClaw's trading capabilities are limited to the available trading scenarios and account system, and funds must be transferred manually by the user. This design intentionally restricts agent capabilities to a verifiable, understandable, and manageable scope, eliminating the uncertainty of over-provisioned permissions.
All GetClaw execution operates within mechanisms that can be isolated, limited, and audited. Users retain great control over the maximum capital allocated to the agent, the types of trades it can execute, and can review all outcomes at any time. Even in the event of automated trading, risk impact is contained to the dedicated account unit, eliminating spillover effects of strategy anomalies, operational errors, or execution deviations to the user's overall portfolio.
Through these three principles, Bitget is building a robust security foundation for agent trading: allowing agents to participate fully in markets while always operating within clear, trusted, and manageable boundaries.
The world's first agent-native exchange: Bitget's vision for the future of markets
We are witnessing a fundamental paradigm shift: from AI assisting humans in trading to AI participating in markets as an independent actor.
Bitget believes the future of financial markets will be an ecosystem where humans and agents coexist and collaborate over the long term. As agents demonstrate increasing advantages in speed, availability, data processing, and execution discipline, autonomous agents will become a major and permanent class of market participants.
With the launch of GetClaw trading capabilities, Bitget is treating agents as a user type from day one, building account systems, trading interfaces, and interaction models designed explicitly for this new participant. Rather than bolting AI on top of an existing trading flow, we are redefining how AI connects to markets, starting at the most fundamental layers: accounts, permissions, and task execution logic.
As the world's first AI agent-native exchange, Bitget will continue to invest in agent account infrastructure, trading capabilities, task orchestration, and ecosystem integration to accelerate the arrival of the agent-first trading era.
Join Bitget, the World's Leading Crypto Exchange and Web 3 Company
Share
