Every few days, it seems, another trading platform allows clients to tether their AI agents to their trading accounts. Woodstock, a Japanese investing app that follows the Robinhood model, is the latest to join the Model Context Protocol (MCP) craze.
According to a statement, the company has released the Woodstock MCP service to “lower the barrier to investing through AI-assisted support.”
Not All MCPs Are Created Equal
The degree of autonomy for an MCP integration across the industry remains uneven. Most deployments so far have included some form of a walled garden to prevent algorithmic chaos.
For instance, the multi-asset broker Capital.com requires a two-step confirmation process before an agent can execute a trade.
Others, like Robinhood and eToro, have cordoned off specific portfolios to protect a client’s primary capital.
It’s unclear what restrictions Woodstock has imposed, if any.
According to the company, it will allow clients to retrieve current figures for share prices, market capitalisation and price-to-earnings ratios. It can also summarise financial statements and aggregate historical movements for US equities. Beyond mere data retrieval, the tool facilitates market, technical and fundamental analysis.
- Claude Powers Nine of Ten Broker AI Agents That Now Trade Live Accounts
- The return of the travel concierge, this time powered by artificial intelligence
- AI Can Mimic Bloomberg. Replacing the Terminal Is Another Matter.
It can calculate resistance and support lines or prepare portfolio rebalancing proposals, and more importantly, it can place buy and sell orders, including split orders.
Woodstock plans to develop a knowledge base of AI prompts. The company intends to share these resources to help its user base refine their decision-making.
Regulators and Robots
Recent data suggests that retail participation is starting to solidify.
According to eToro, 19% of individual investors currently leverage AI applications for asset selection or portfolio adjustment, a notable rise from 13% just twelve months ago, while 39% express a willingness to adopt such technology.
Should the AI agent become the primary gateway to financial markets, the implications are stark. The trading app will most likely become a data and execution pipeline. The full scope of this evolution, from destination to utility, is still unfolding.
Less certain is the fallout when an AI agent goes rogue. Regulators have yet to provide clear signals, if any at all.
The EU AI Act contains a "human-in-the-loop" provision, which provides at least one direction. This requirement for human oversight likely explains the restrictions seen at Capital.com and elsewhere.
But if the AI agent becomes the de facto trader, the industry must ask whether existing MiFID rules are fit for purpose.
Rules designed around human behaviour and traditional risk profiles may struggle when faced with a machine that never sleeps and never doubts.
