Madgwick update: MCPs connect AI to tools
LabNext70's Madgwick update introduces MCP (Model Context Protocol) tools that fundamentally change how AI assistants work.
MCPs are secure connections that allow AI models to interact with external applications, tools, and data sources. Think of them as AI-specific plugins that transform Madgwick from a chat interface into a comprehensive workspace where assistants can search databases, create images, analyse data, and perform specialised tasks without leaving the platform.
Where to find MCPs
The MCP Store serves as your central hub for discovering and installing these tools. Currently in trial phase, the store is being populated gradually to ensure stability and proper testing.
Click on your initials in the top right, and the MCP Store link at the top of the drop down, or use the Box icon in the bottom left of your chatbox and click Manage Connections.
This opens the MCP Store popup, which includes search and filtering options.
MCP Tool installation process
Browse available MCP tools by category
Click "Install" on your selected MCP tool
Click 'Add Connection'
The MCP appears in your available MCP tools list
See our walkthrough webinar and UI PDF and MCP guide documentation for more detail.
MCP Guide
Practical applications
MCPs enable workflows that would previously require juggling multiple tools. The platform can now handle complex, multi-step tasks within a single conversation thread.
For example, researching a topic might involve Academic Search pulling recent papers, Firecrawl extracting data from relevant websites, a search of relevant X posts by X Deep Search, and Perplexity providing comprehensive background - all coordinated through a single assistant that synthesises the findings.
What to expect during the trial
In this trial phase new MCPs are being released gradually to ensure stability.
Trial phase considerations:
Some models work better with MCPs than others (see Madgwick’s new Model Catalogue for a guide)
Since MCPs connect to third-party services, connection issues or API changes can temporarily affect functionality.
Occasional connection failures – try refreshing the page if connections hang
Watch the LabNext70 newsletter for MCP release info
Send us your feedback to improve the platform