From Prompt to Production: Advanced Gemini CLI for Serverless

The Gemini CLI brings Google's powerful AI models directly to your terminal - no context switching, no web UI required. But this isn't an intro session. Dan Dobrin returns to ServerlessTO to reveal production-ready patterns that move beyond basics: how to engineer AI memory, automate your entire development workflow, and orchestrate multi-agent systems - all from your command line.

Through hands-on demonstrations with MCP (Model Context Protocol) integrations and serverless deployments, you'll see how terminal-first AI workflows can transform everything from code reviews to production deployment. This is practical, production-ready implementation - not theoretical concepts.

Resources (Dan will touch upon them during the session):

• Gemini CLI "gemini-plan-commands" extension https://github.com/ddobrin/gemini-plan-commands
• Gemini CLI's Hidden Superpower: Custom System Prompts - by Jamie Duncan https://jduncan.io/blog/2025-11-13-gemini-cli-custom-system-prompts/
• Gemini CLI tips https://github.com/addyosmani/gemini-cli-tips
• Gemini CLI cheatsheet https://www.philschmid.de/gemini-cli-cheatsheet
• Cloud Run MCP server https://github.com/GoogleCloudPlatform/cloud-run-mcp
• Spring AI - Google GenAI SDK Integration 1.1.0 GA https://spring.io/blog/2025/11/12/spring-ai-1-1-GA-released#new-provider-integrations
• Google GenAI Java SDK https://github.com/googleapis/java-genai
• Gemini CLI https://geminicli.com/

What You'll Learn

• Ever wonder why AI assistants get confused or forget what you've told them? We dive into context engineering and understand short vs long-term memory
• What can I shift left in our work? Could it be code reviews, documentation, security, or building pipelines?
• Are there any tools available to us, any extensions we can leverage? Exploring the ecosystem
• MCP is all the rage - What is available to us when building and deploying a Serverless App
• Let's plan before we build - Following the path of analysis, planning, refinement, implementation and deployment
• Building multi-agent systems with Gemini CLI? Orchestrating multiple agents working together

Agenda

• 12:00 - 12:10 PM: Welcome and community updates
• 12:10 - 12:50 PM: Main presentation with live demonstrations
• 12:50 - 1:20 PM: Q&A and discussion
• 1:20 - 1:30 PM: Manning Publications raffle

About the Speaker
Dan Dobrin is an Enterprise Application Architect at Google Cloud with extensive experience in enterprise Java applications, cloud technologies and generative AI. He recently presented multiple sessions at Google Cloud Next '25, Spring IO 2025, focusing on Agentic AI, Spring AI, LangChain4j, Gemini Code Assist, Gemini CLI and Agentic RAG systems. Dan previously spoke at Serverless Toronto about High-Performance Java on Google Cloud Platform, and Building Agentic AI systems: A2A (Agent2Agent) & MCP Protocols & Serverless and is a frequent speaker at conferences. With a focus on practical, production-ready implementations, Dan brings real-world insights into cutting-edge technologies. His most recent work focuses on contributing to the Google GenAI SDK to Spring AI and Google Agent Development Kit.

Who Should Attend

• Developers and engineers building apps for serverless environments, who want to streamline their workflow, test prompts quickly, review code, analyze app security or integrate AI capabilities into their command-line tools and scripts
• DevOps and infrastructure teams who deploy to Serverless environments could benefit from automating tasks, generating scripts, or troubleshooting issues directly from the terminal without switching contexts to a web interface
• Data scientists and ML engineers who want a lightweight way to experiment with Gemini models, prototype AI features, or process data using AI without heavy IDE overhead
• Technical writers and documentation teams who need to generate, refine, or analyze content at scale and would benefit from scripting and automation capabilities that a CLI provides
• System administrators looking to leverage AI for log analysis, system diagnostics, or automating repetitive tasks directly from their terminal environment
• Power users and productivity enthusiasts who prefer keyboard-driven workflows and want to integrate AI assistance seamlessly into their existing terminal-based toolchains

The talk would be especially valuable for anyone who spends significant time in the terminal and wants to avoid context-switching to web UIs while maintaining access to powerful AI capabilities.

Format: Virtual lunch & learn (12:00-1:30 PM EST). Grab your lunch and join us!