Скачать или смотреть FDC3 and MCP: The Standard That Makes AI Deterministic in Banking

In this practical workshop from the Open Source in Finance Forum (OSFF), Robert Myers and Kalin Kostov from interop.io demonstrate how to turn the "future of AI" into a "right now" reality by mapping FDC3 Intents to the Model Context Protocol (MCP).

🔑 FDC3 and AI: Using FDC3 Intents as MCP Tools | interop.io Workshop

🇬🇧 Build the AI Desktop in London! Join the leaders of the AI-interop movement at OSFF London on June 25, 2026, at the Queen Elizabeth II Centre. 🎟️ Register Now: https://hubs.ly/Q041YV9Z0 🔥 20% OFF CODE: 26YTOSFFLN20C

🚀 Explore FDC3: https://fdc3.finos.org/
🌐 More about FINOS: https://www.finos.org/
📧 Join our newsletter: https://www.finos.org/sign-up
📥 Download the State of Open Source in Financial Services report: https://www.finos.org/state-of-open-s...
🎙️ Listen to our Open Source in Finance Podcast:    / @finos  
🗣️ Attend the next Open Source in Finance Forum: https://hubs.ly/Q03z9D9D0
LinkedIn:   / finosfoundation  

🤖 Making AI Deterministic and Action-Oriented
While LLMs are great at chatting, they often struggle to "do" things within a secure financial environment. Bob and Kalin show how FDC3 Intents—the standard for cross-app actions like "ViewChart" or "StartCall"—provide the perfect, secure foundation for AI tools.

FDC3 as a "Toolbox": Instead of letting an AI guess how to interact with a trading terminal, you expose FDC3 intents as MCP (Model Context Protocol) tools. This allows the AI to trigger specific, authorized actions across your existing apps.

Minimizing Data Requests: By using intents, the AI doesn't need to ingest your entire database; it simply sends a standardized "intent" to the app that already has the data.

Deterministic Results: This mapping ensures that when an AI "takes action," it does so through a community-vetted, secure standard, preventing hallucinations in execution.

🏗️ Workshop Highlights: From Prompts to Production
The session moves past theory into a hands-on guide for financial engineers:

Prompt Engineering for FDC3: How to craft high-quality prompts that help LLMs select the right FDC3 intent for a user’s request.

Overcoming Metadata Challenges: Solving for schema definitions and discoverability so your AI "knows" what tools are available on the user's desktop.

Real-World Integration: Tips for selecting LLMs and chat UIs that work within the strict security constraints of a bank or hedge fund.

⚙️ Why This is the "Right Now" Strategy
Leverage Existing Tech: You don't need new apps; you just need to expose your current FDC3-enabled apps to your AI agents.

Safety First: Maintain the "human-in-the-loop" security that FDC3 app agents provide.

Efficiency: Drastically reduce the time it takes to build "Copilots" by using open-source standards as the connective tissue.

The takeaway: FDC3 intents are the "digital hands" that allow AI to move from being a passive observer to an active participant in the financial desktop. By using MCP as the bridge, we are creating a secure, scalable way for AI to orchestrate the most complex workflows in finance. See these "Agentic" workflows live at OSFF London on June 25!

#FINOS #OSFFLondon #FDC3 #AI #MCP #interopio #ModelContext Protocol #FinTech #GenerativeAI #AgenticAI #OpenSource #SoftwareArchitecture