Your AI assistant can now prepare trades for you. Brokers just made it official

For years, AI in retail trading meant chatbots that answered support questions and signals that promised more than they delivered. In the last week of September, that changed. On 29 September, Webull expanded its cloud server for the Model Context Protocol, a standard that lets AI assistants connect to outside software, so eligible US clients can use mainstream AI apps to research markets, analyse portfolios and prepare trades, without coding or installing anything, according to Finance Magnates. The same day, prop firm The5ers let traders connect popular AI chat apps to monitor their accounts. Two days later, platform provider Match-Trader released a management connector for brokers and prop firms with more than 20 read-only actions. Three announcements in one week mark the moment AI stopped being a feature and became an interface.

How AI-prepared trades actually work

The important detail is what the AI cannot do. Webull's assistant does not send orders to the market. It builds the order instruction and passes it to Webull, where the customer must review and confirm it in the broker's own app or desktop platform. The human stays in control of the final step, which keeps the broker within existing rules on order authorisation and suitability.

Before that step, the assistant can do a great deal. According to Webull's announcement, users can pull real-time and historical data across stocks, ETFs, options, futures, crypto and event contracts, research company fundamentals and filings, run screeners and view balances and positions. In practice, a client can ask in plain language for the five largest positions with earnings this week, a summary of risk, and a draft order to trim one of them.

That changes the experience of using a broker. Instead of navigating menus, screens and order tickets, a client describes what they want and checks the result. For experienced traders, it saves time. For newer traders, it lowers the barrier to doing research that previously required skill with complex platforms.

The service is available to eligible users in the United States through several supported AI platforms, with more pending approval. That limited roll-out reflects regulatory caution: brokers want to see how clients use the tools before expanding them to more markets.

Prop firms and platforms are moving even faster

The5ers, a well-known prop firm, integrated mainstream AI chat apps for account monitoring through the same protocol, Finance Magnates reported on 29 September. A funded trader can ask an assistant how close they are to their daily loss limit, how their drawdown compares with the rules, or which trades hurt their performance this month. For prop traders, whose accounts can be lost by breaking a single rule, real-time answers in plain language are genuinely valuable.

Match-Trader went a step further for the industry side. On 1 October it gave brokers and prop firms a management connector offering more than 20 read-only actions. That allows a firm's own staff to query client accounts, exposure and activity through an AI assistant, which could speed up risk monitoring, support and reporting. Read-only is the key phrase: the assistant can see but not change.

The pattern is consistent across all three. AI gets broad access to read data and limited or no power to act. That is a sensible starting point for a regulated industry, and it reflects lessons learned from earlier automation, where errors in automated trading caused large losses before anyone noticed.

The trend builds on moves we covered earlier in the year, including MetaQuotes adding AI automation to MT5. Platform providers, brokers and prop firms are converging on the same idea: the AI assistant becomes a front door to the trading account.

• Webull, 29 September: AI assistants can research, analyse and prepare orders; clients must confirm
• Webull data access: stocks, ETFs, options, futures, crypto and event contracts
• The5ers, 29 September: AI chat apps can monitor prop accounts via MCP
• Match-Trader, 1 October: management connector with 20-plus read-only actions
• Common rule: broad read access, limited or no power to execute

The risks brokers must manage

Giving an AI assistant access to account data raises obvious questions about privacy and security. The data flows from the broker to an external AI platform, which may store or process it under its own terms. Brokers need clear agreements on how that data is handled, and clients need to understand what they are sharing. A breach involving trading account data would be serious for any firm.

Accuracy is the second risk. AI assistants can misread instructions or produce confident but wrong answers. If an assistant prepares an order with the wrong quantity or instrument and a client confirms it without checking, who is responsible? Requiring confirmation helps, but brokers will need clear warnings, sensible defaults and safeguards such as order size limits to reduce errors.

Suitability and advice rules are the third. If an assistant suggests trades, regulators may ask whether that amounts to personalised advice, which requires licensing in most markets. Brokers will want assistants to help clients act on their own decisions rather than recommend trades, a line that will need careful design and documentation.

Finally, there is fraud. Scammers already impersonate brokers and support staff. As AI assistants become a normal way to interact with accounts, criminals will try to trick clients into connecting fake tools or approving orders they did not intend. Education will be essential.

What it means for brokers in Southeast Asia

Southeast Asian traders are among the most mobile-first in the world and quick to adopt new apps. Once AI-prepared trades become common in the United States, clients in Singapore, Bangkok, Jakarta and Manila will expect the same. Brokers that can offer AI access in local languages, with clear safeguards, will have an advantage in attracting younger, digitally active clients.

Regulators in the region are also thinking about AI in finance. Singapore's central bank has published guidance on AI risk management, which we discussed in our report on MAS and AI agents. Brokers planning AI features should build them with those expectations in mind rather than retrofitting controls later.

Language support will decide adoption in the region. Many traders in Thailand, Vietnam and Indonesia prefer to work in their own language, and an assistant that answers only in English will reach a limited audience. Brokers that test assistants carefully in Thai, Vietnamese, Bahasa Indonesia and Tagalog, and confirm that order details are interpreted correctly, will reduce errors and build trust faster than those that switch features on without local testing.

The AI assistant is becoming the front door to the trading account.

How brokers should prepare

Brokers do not need to rush to launch, but they do need a plan. That means deciding which data an assistant can access, what actions it can prepare, how confirmation works, how data is protected and how clients are educated. It also means talking to platform providers about connectors, since many brokers will get AI access through their trading platform rather than building it themselves.

Staff training matters as much as client education. Support teams will receive questions about AI-prepared orders, data sharing and errors, and they need clear answers. Compliance teams need to understand how the tools work in order to supervise them. A broker that launches AI access without preparing its own people risks confused clients and inconsistent responses when something goes wrong.

Finally, brokers should measure results honestly. Tracking how many AI-prepared orders are confirmed, edited or abandoned, and how often clients report errors, will show whether the feature helps or harms. The firms that learn fastest from real usage will shape the standards others follow.

The first week of AI-prepared trades will not transform the industry overnight. But it marks a clear shift in how clients will interact with brokers. The menus and order tickets that defined online trading for two decades are about to become something clients rarely touch. Brokers that design for that change carefully, with safety built in, will win the next generation of traders. Those that ignore it will find their platforms feeling old faster than they expect.