Beyond the IDE: A new era of agent collaboration

What if your AI coding partner could learn from its mistakes and update its own workflow to become more efficient over time?

#1about 3 minutes

Grappling with the uncomfortable truths and motivations of AI

AI's impact on jobs raises questions about its purpose, which ranges from reducing toil and breaking barriers to pure capitalism.

#2about 3 minutes

Introducing Gemini CLI for direct model interaction

Gemini CLI provides lightweight, bare-metal access to models, grounding them with local files and search while enabling automation and personalization.

#3about 4 minutes

Live demo of an AI agent fixing a GitHub issue

A live demonstration shows how to prompt Gemini CLI to plan a fix for a real GitHub issue, utilizing an MCP server and a ReAct loop.

#4about 4 minutes

Guiding AI agent behavior with structured markdown files

The AI agent's workflow is customized by combining three types of markdown files for personal process, team best practices, and project structure.

#5about 4 minutes

Observing the AI agent execute a test-driven development plan

The AI agent follows a test-driven development plan by writing a failing test, implementing the fix, and troubleshooting errors until the tests pass.

#6about 2 minutes

Improving the AI agent's process through reflection

By prompting the agent to reflect on its troubleshooting session, it generates improvements that are then saved back into its guiding markdown files.

#7about 3 minutes

The strategic choice of TypeScript for Gemini CLI

Gemini CLI was built with TypeScript to maximize community contribution, leverage team familiarity, utilize the rich ecosystem, and ensure portability across environments.

#8about 3 minutes

Exploring diverse real-world use cases for AI agents

Real-world applications of Gemini CLI range from SRE root cause analysis at Delivery Hero to personal knowledge management and solving philosophical logic puzzles.

#9about 2 minutes