From portfolio construction and risk analysis to benchmark comparisons and order preparation – all within a single AI conversation.
The Investment Decision Lab now offers a live Model Context Protocol (MCP) server, enabling AI assistants such as ChatGPT and Claude to interact directly with its portfolio construction and analytical capabilities.
Rather than navigating individual functions through a traditional user interface, investors can explore portfolios, challenge assumptions, evaluate alternative allocations and request detailed analyses and reports through natural-language conversations.
Connecting Investment Analytics with Market Data
The real potential of MCP becomes apparent when multiple services are connected to the same AI assistant.
In recent tests, the Investment Decision Lab was combined with the MCP server of market data provider EODHD, enabling workflows that go beyond the Lab's standalone functionality.
Two examples illustrate the possibilities:
1. Historical Benchmark Comparison
An existing portfolio was compared against a benchmark consisting of 80% global equities (MSCI ACWI) and 20% Swiss government bonds.
Using historical ETF prices, the analysis calculated annualised returns, volatility, maximum drawdowns, correlations and relative performance.
Over the period from March 2018 to October 2026, the portfolio achieved an annualised return of 8.2%, compared with 7.3% for the benchmark, while exhibiting lower volatility (11.3% versus 12.5%).
2. Portfolio Order Preparation
The AI assistant also generated an executable-style order list for a CHF 100,000 investment based on the Lab's target allocation.
By retrieving market prices through EODHD, the analysis determined ETF quantities, estimated order values, currency conversions and residual cash.
The output included security identifiers, exchange-specific tickers and pricing timestamps, together with warnings about potentially stale or inconsistent market data.
These examples demonstrate how an AI assistant can combine specialised financial models with external data sources in a single workflow.
Towards Integrated Investment Workflows
The MCP architecture is not limited to market data providers.
Additional MCP servers can connect brokerage platforms and other financial services, potentially extending the process from portfolio construction and simulation to order preparation, review and approval.
Importantly, these integrations do not need to be built into the Investment Decision Lab itself. The AI assistant orchestrates the interaction between independent services.
This creates a new level of flexibility for professional investors – while preserving the Lab's transparent, quantitative investment methodology.
The underlying investment calculations remain deterministic. AI provides the conversational interface and coordinates the workflows.
Appropriate controls, data validation, investor suitability checks and human approval remain essential, particularly where investment decisions lead to transactions.
From Software Application to Connected Investment Infrastructure
Until recently, such integrated workflows would have required substantial custom development.
With MCP, independently developed applications can now work together through a common interface, allowing existing investment tools to become components of broader AI-assisted processes.
The Investment Decision Lab demonstrates this approach in practice.
Explore the Investment Decision Lab: https://www.bicon.co
The Investment Decision Lab MCP server is live. External market data and brokerage integrations depend on separately available MCP servers and are not included in the Lab's own functionality. The examples shown are analytical demonstrations, not executed transactions.
Disclaimer: The examples are for demonstration purposes only and do not constitute investment advice. Market data may be delayed or incomplete. External MCP integrations are provided by third parties and are not part of the Investment Decision Lab. All analyses and orders require independent verification and appropriate approval before execution.