Skip to Content
View all events

MCP with Local AI Models Using LM Studio and Goose Desktop

Published by O'Reilly Media, Inc.

Intermediate content levelIntermediate

Creating agents without sending data to the cloud

What you’ll learn and how you can apply it

  • Select the optimal local models for MCP tool-calling based on your hardware
  • Learn how to use LM Studio and Goose Desktop as a drop-in replacement for ChatGPT and Claude
  • Deploy LM Studio as a local server for REST API calls with MCP server support
  • Deploy Goose Desktop as a local AI agent framework with Ollama

Course description

Large language models (LLM) have powerful capabilities that are able to reduce the burden of work for countless tasks across industries, and the Model Context Protocol (MCP) allows developers to create AI agents that have access to custom data sources to augment the knowledge of traditional LLMs. In this course, you’ll learn how to use both LM Studio and Goose Desktop as drop-in replacements for ChatGPT and Claude. You’ll understand how to configure both applications with freely available small LLMs that you can download and configure on your desktop or local server for private, on-premises deployments.

You’ll also discover how to leverage offline LLMs as a server and migrate AI agents that use ChatGPT APIs to invoke the REST API of your offline model. This enables you to build 100% offline AI-powered applications, with the tool-calling support of MCP, but without paying per million tokens to AI service providers. Local models with API REST services are perfect for developers in regulated industries or anyone prioritizing privacy, security, and cost control. With Bruce Hopkins’ guidance, you’ll gain practical experience configuring MCP servers, connecting local AI models, building production APIs, and creating agents that can access files, databases, APIs, and external tools without sending data to the cloud.

This live event is for you because...

  • You’re a developer who needs to build AI agents that perform real tasks.
  • You’re an AI engineer looking to streamline agent development and reduce time to production.
  • You’re a technical lead evaluating platforms for building enterprise AI applications.
  • You’re familiar with ChatGPT/Claude but want to create custom agents that access your organization’s data.
  • You want to future-proof your skills as AI agents become integral to modern applications.

Prerequisites

  • Python 3.10+ installed on your machine
  • Basic command-line proficiency
  • An understanding of REST APIs and JSON
  • Basic programming experience in any major programming language like Python or JavaScript
  • Familiarity with HTTP APIs and JSON
  • Experience as a user with ChatGPT, Claude, or other LLMs
  • Understanding of basic software development concepts (functions, variables, APIs)

Schedule

The time frames are only estimates and may vary according to how the class is progressing.

MCP fundamentals (60 minutes)

  • Presentation: Understanding what MCP is and why it matters; differentiating between MCP clients, servers, and transports; connecting an MCP server to Claude Desktop; tool-calling in real time
  • Q&A

The case for local LLMs for on-prem deployments (60 minutes)

  • Presentation: Understanding security and compliance benefits of local AI; evaluating cost implications of cloud versus local models; identifying use cases where local models excel
  • Break

Getting started with LM Studio and MCP (60 minutes)

  • Presentation and demonstration: Installing and configuring LM Studio; downloading and loading optimal models for tool-calling; connecting MCP servers to LM Studio; building a multitool AI agent running entirely offline
  • Q&A

Using Goose Desktop for agentic workflows (55 minutes)

  • Presentation and demonstration: Implementing tools for app search and discovery; installing Ollama as the model runtime; configuring Goose Desktop with local models; adding custom MCP extensions to Goose; building autonomous file system and API agents

Wrap-up (5 minutes)

  • Presentation: Ideas for extending the app directory and creating new apps; resources for continued learning; community engagement opportunities
  • Q&A

Your Instructor

  • Bruce Hopkins

    Bruce Hopkins is a technical writer, an AI expert, an Intel Software Innovator for AI, and an Oracle Java Champion. He’s also the author of ChatGPT for Java and the coauthor of Beginning ChatGPT for Python.

    Xlinksearch

Skills covered

  • MLOps
  • ML.NET
  • GPT
  • AutoML