ThinkMarkets has launched a Model Context Protocol (MCP) server, ChelseaAI, which allows access to its trading platform via any AI client used by traders, Nauman Anees, co-founder and CEO of the broker, revealed to Finance Magnates. “Traders can now connect any AI LLM of their choice and use it to place trades without ever logging into the trading platform,” he said, adding that “It will change the way traders ideate and make trading decisions.”

“This means you no longer need to worry about charting, automated trading, indicators, or analysing market data. Your AI handles all of it,” he added. “That alone removes a tremendous amount of user friction.”

“AI Cannot Access Traders’ Funds…but Can Execute Orders”

However, there are limitations when it comes to access to funds, which ThinkMarkets stressed “is by design and not by policy.”

Anees explained that “AI cannot access traders’ funds or make deposits or withdrawals,” but it can execute orders. ThinkMarkets also created something called scopes, which allows traders to set permissions for AI on whether it can execute orders or not.

“You have full control over which order types are scoped within your AI platform,” the CEO continued. “Looking ahead, ChelseaAI will be a complete trading assistant from analysis, ideation, account management to execution, making her less of a tool and more of an AI companion that can execute your entire trading vision.”

Read more: Revolut Built a Trading Desk With AI in 30 Minutes. Will Prompts Replace Broker Platforms?

To have premium access, ThinkMarkets customers will receive AI token credits as incentives from the platform. These incentives will be tied to trading activities.

Overreliance on AI, meanwhile, can also expose traders to risks. As agentic functions are allowed, the broker must be cautious about the safety limits for client funds.

“The AI can't execute your entire account on a single trade or margin. That's by design,” Anees added. “There are circuit breakers and hard limits in place, but beyond that, you also have granular control over what happens within your chats. You can tell ChelseaAI to focus purely on analysis without executing any orders, and it will respect that completely.”

“As for the risk of the AI going haywire, that's exactly why we built a permissions and scopes system. You can revoke the AI's execution permissions entirely, meaning even in a worst-case scenario, it can never access or move your funds.”

You may also like: Is Jefferies Preparing to Offload FXCM?

ChelseaAI Is Not Supported for Third-Party Platforms as They Are “Not Our Platforms”

ThinkMarkets’ MCP server can be connected to any AI assistant used by the trader to link it with ThinkTrader, which is ThinkMarkets’ own trading platform. Although the broker offers trading on other platforms, Anees said that ChelseaAI is not supported on them as they are “not our platforms.”

“We can control a lot of the permissions and scopes on our proprietary platforms,” he said. “That's not something we can do, obviously, on other external platforms.”

Headquartered in Melbourne and London, ThinkMarkets operates globally with licences in those countries, along with Cyprus, South Africa, Dubai, New Zealand, and a few offshore jurisdictions. It also established a presence in Japan after acquiring a local brokerage in 2021.

The new AI MCP server is going to be available to all ThinkMarkets clients globally. Although the experience appears streamlined on the surface, the question of data security arises, especially in Europe, where GDPR is prevalent.

“You share your data, but you still control your data,” Anees stressed. “Your data never leaves the account ecosystem. So every time you close a session, we don't keep track of any of that. That's between you and the LLM.”

Although LLMs can access trading-related data, ThinkMarkets “does not see any regulatory friction” around it, as the user would still control their data. “If we move the data, then it's a separate topic,” said Anees, “and we never do that.”

“The Value of an Interface Is No Longer as Critical as the Value of the Underlying Product Features”

The CFD trading industry is dominated by MetaTrader, but over the years, many other trading platforms have evolved. Big brokers like ThinkMarkets have also developed their own trading platforms, which require a lot of resources and time but give them more control. The arrival of MCP servers now raises a big question: will it decrease the incentive for brokers and other trading platform providers to improve their interfaces?

“What we need to realise is that with APIs, MCPs, cloud infrastructure, and GPTs, the value of an interface is no longer as critical as the value of the underlying product features. The two are worth separating,” the ThinkMarkets CEO said.

“You can design beautiful interfaces indefinitely, but the reality is that a significant amount of trading today happens through APIs. Manual trading has its place, but it also has its limits. If you want to sit in front of a chart all day, that's great. But now you don't have to. AI can analyse the chart for you, handle the heavy lifting, and even help you learn along the way. The interface stops being a barrier. Your trading life becomes simpler, and your focus can shift to what actually matters, your strategy and your vision.”

BREAKING: We just gave Claude access to the entire options and stock market and it's not a demo. It's the Unusual Whales MCP Server. It plugs directly into any AI assistant and gives it live, structured market data on demand. Build a trading bot. Build a finance dashboard.… pic.twitter.com/XKL554g7f2

— unusual_whales (@unusual_whales) March 11, 2026

ThinkMarkets, meanwhile, is not the first in the brokerage industry to launch an MCP server for traders. Recently, Spotware, the developer of cTrader, also launched a similar MCP server, while eToro also has something called “Agent Portfolios” to let users link their own AI agents to live trading accounts. While the Aussie division of IG Group also launched a "read-only" MCP server, limiting its access to ChatGPT for now and with plans to allow Claude "soon", Robinhood went all-in, enabling AI agents to trade stocks and make payments.

The latest launch of ChelseaAI by ThinkMarkets also indicates that it is becoming a trend among brokers and tech providers to provide MCP server access to traders.

Anees thinks that a lot of industry players “will now attempt” to launch such MCP servers. “But making an AI integration to an MCP server is not something that can be done instantly. And it has to be built with your own technology. So many can try and offer it, but we're one of the few that have done a deep integration,” he continued. “That's why we keep your data secure. That's why we have circuit breakers and controls. And that's why we have designed it truly for agentic trading. So just offering it is not going to be enough.”

This article was written by Arnab Shome at www.financemagnates.com.