Coding and Personal Agents with Ollama | LIVE145

Michael Chiang walks through using Ollama to run open models locally and in the cloud, and how to wire those models into real applications. The session also demonstrates setting up Copilot CLI and connecting an agent to tools like web search and vision, with a focus on hybrid patterns and privacy.

Overview

A live walkthrough showing how developers can:

• Launch and use Ollama for running open models locally.
• Switch between local and cloud model execution as part of a hybrid AI architecture.
• Set up Copilot CLI and use its autonomous capabilities for code exploration, planning, and execution.
• Connect an agent to external tools such as:
  • Web search
  • Vision (image understanding)
• Keep sensitive data local to improve privacy and control while still using cloud models when needed.

Session chapters (from the video description)

• 0:00 - Fun fact: O Llama mascot designed by Michael’s wife
• 00:04:03 - Introducing seamless experience between local and cloud model usage
• 00:06:00 - Copilot CLI’s autonomous code exploration, planning, and execution features
• 00:06:51 - Discussion on hybrid AI architecture: balancing local and cloud operations
• 00:08:18 - Handling sensitive data locally for greater privacy and control
• 00:08:41 - Use of coding agents and open models through cloud for development efficiency
• 00:11:17 - Developer Experience and Product Design Philosophy
• 00:12:56 - Explaining the Simplicity and Memorability of the Logo
• 00:13:27 - Affirmation of O Llama’s Logo as Easily Recognizable