Webinar Summary
The following summary is auto-generated from the webinar recording.
In the rapidly evolving world of AI and IoT, developers are constantly searching for tools that simplify integration and accelerate innovation. One such breakthrough is the use of Model Context Protocol (MCP) servers, which serve as a bridge between large language models (LLMs) and real-world IoT devices and data. This webinar dives deep into how MCP servers are transforming IoT AI development, focusing on Blues’ innovative MCP servers and how they empowers developers to harness AI insights effortlessly.
Understanding MCP Servers and Their Role in IoT AI Development
MCP servers act as an open, standardized protocol that connects applications with large language models. They enable LLMs to query tools, resources, and prompts exposed by MCPs to access real-world data and perform actions. Think of MCPs as the "USB-C port" for AI applications, providing a universal interface that allows AI models to interact seamlessly with various data sources and services.
The recent explosion of MCP servers, pioneered by companies like Anthropic and adopted by platforms such as OpenAI's ChatGPT, VS Code, and Cursor, has opened up new possibilities for integrating AI with IoT devices. By providing both context and documentation baked into their APIs, MCPs ensure that LLMs understand not just how to call an API, but also how to use it effectively, reducing guesswork and hallucination risks.
The Blues Notecard and Notehub Ecosystem
At the heart of the Blues IoT ecosystem is the Notecard—a wireless connectivity module designed to simplify cloud enablement especially in cellular IoT. Unlike a simple modem, the Notecard contains radios, circuitry, and firmware that allow it to function as a standalone module connecting products to the cloud with just two lines of code. Notecards offer global cellular coverage across 140+ countries, prepaid data plans with no monthly SIM fees, and low power consumption, making them ideal for IoT deployments.
Complementing the Notecard is Notehub, Blues’ cloud service that securely syncs data bi-directionally between devices and cloud applications without requiring complex configuration. This seamless integration means developers can focus on product innovation without the hassle of managing connectivity layers.
How MCP Servers Enhance Blues IoT Development
Blues has developed three primary MCP servers designed to accelerate your developer journey by providing an AI-driven interface to their products:
- Notecard MCP: Runs locally and abstracts the Notecard CLI, allowing LLMs to query and control Notecards connected to your machine.
- Notehub MCP: Acts as a wrapper around the Notehub API, providing context-aware access to fleet management, device data, and project information.
- Developer MCP: Focuses on best practices and nuanced guidance for building firmware and applications with Blues tools, helping to reduce hallucinations and errors during development.
These MCPs not only expose APIs but also embed rich metadata and documentation, enabling LLMs to make intelligent decisions about which tools to use and how to chain them together for complex workflows.
MCP Server Tools, Resources, and Prompts
When building MCP servers, three components are critical:
- Tools: Functions or actions that the MCP exposes. LLMs can autonomously decide to use these based on context without explicit instructions. For example, sending a Notefile from a Notecard to Notehub.
- Resources: Static or dynamic data sets such as API documentation, markdown files, or datasets that the LLM can reference. Blues uses resources to provide API endpoints and best practice documents directly within the MCP environment.
- Prompts: Predefined instructions that shape how the LLM behaves before interacting with tools. While Blues hasn’t found prompts especially useful in their MCPs, they can be used to configure the interaction based on user expertise or preferences.
Embedding comprehensive documentation and context into MCPs is vital because it teaches the LLM how to use the APIs effectively, much like teaching a new user to navigate tools.
Connecting and Managing Notecards with Claude Desktop
Claude Desktop, a simple and intuitive desktop application for Mac and Windows, serves as an excellent MCP client for interacting with Blues’ MCP servers. Through Claude, developers can connect to a Notecard over USB, query device details, check firmware versions, and perform updates—all via natural language commands.
For instance, by asking Claude, “Tell me about my Notecard,” the MCP server uses the Notecard CLI in the background to retrieve device ID, firmware build, board version, and capabilities like Wi-Fi, cellular, and GPS. It can also check for firmware updates and apply them seamlessly.
Developers are prompted for permissions before the LLM uses any MCP tool, ensuring security and control over local executables. This interaction highlights the balance between automation and user oversight that MCPs provide.
Configuring Wi-Fi and Provisioning Devices
Claude can instruct the Notecard to connect to a Wi-Fi network using simple commands, such as providing an SSID and password. It then confirms successful connection using Notecard requests. Further, connecting a Notecard to a Notehub project can be done by simply asking Claude to provision the device to a specified project. The MCP servers handle all the underlying API calls and synchronization, abstracting away complex workflows.
Visualizing Device Fleets with Notehub MCP
Blues’ Notehub MCP server enables natural language queries against device fleets. By providing a project UID to Claude, developers or even non-technical stakeholders can retrieve summaries of products, fleets, and devices. The MCP can generate tables showing device locations and last transmission times, or create visualizations like bar charts and pie charts to represent device distributions across countries.
This capability is invaluable for fleet management, allowing quick insights without writing complex queries or code.
Building Firmware with VS Code and Developer MCP
The developer MCP server integrates with tools like VS Code Copilot to assist engineers in writing firmware that follows Blues’ best practices. By prompting the LLM with project requirements, such as building an Arduino project for a Swan MCU with temperature and humidity sensors, the MCP guides code generation with embedded templates and best practice knowledge.
The generated code includes helpful artifacts like flow diagrams (using Mermaid syntax) to visualize logic, README files, and build scripts. Although LLMs can sometimes be verbose or produce errors, the MCP’s guardrails help reduce hallucinations and provide iterative feedback to fix issues like compilation errors.
Once compiled, firmware can be uploaded to development boards like the Swan via an ST link, completing the end-to-end development cycle enhanced by MCP servers.
Challenges and Future Directions
While MCP servers offer tremendous potential, there are challenges such as managing token limits in LLMs, avoiding hallucinations, and handling multiple MCP servers simultaneously. Developers must carefully document tools and resources to minimize confusion and errors.
Blues is also exploring ways to integrate MCP functionality into web terminals and documentation portals, aiming to provide easier access and learning paths without requiring local hardware. The vision includes marketplaces for MCP servers to simplify installation and configuration across tools.
Conclusion
MCP servers are revolutionizing how IoT developers leverage AI by creating intelligent, context-aware interfaces to hardware and cloud services. Blues’ MCP servers—Notecard, Notehub, and Developer MCP—demonstrate how AI can automate complex tasks, provide real-time insights, and guide developers through best practices.
By embedding rich documentation and context, MCPs reduce the friction of working with LLMs and IoT data, transforming the development experience into one that feels collaborative and insightful. As these tools mature, they promise to become indispensable companions in the AI-ready IoT landscape.
For more information and to join the conversation, visit the Blues Community Forum and explore the open-source MCP servers on GitHub.
