collaborating with an AI on complex tasks. Welcome to Software 3.0 As AI becomes a commodity, the focus shifts from custom models to intelligent integration—a new paradigm called Software 3.0, where the LLM is the new operating system. We program with natural language, orchestrating AI agents to solve complex problems. This requires mastering multi-agent patterns, as covered in this great set of posts and in the Agent factory 2nd episode. The Power of Context and Memory The art of AI development is no longer just in the code, but in context engineering—providing the right information at the right time. A crucial part of this is giving our agents memory, and that's why the recent launch of Vertex AI Memory Bank is so significant. As we dive deeper into this topic in the Agent Factory 4th episode, memory is what elevates a simple tool into a truly helpful assistant. This journey is the new standard—challenging, exciting, and we're here to help you master it. Watch on-demand Your Complete AI Journey: From Build to Scale Accelerate AI with cloud run We've designed a complete developer journey to guide you from idea to production across four stages: Inspire, Build, Deploy, and Scale. Your path starts with the Agentverse, an "epic quest" covering the Inspire and Build stages and then transitions to Accelerate AI with Cloud Run to focus on deployment and scaling. Join the Agentverse to face the epic final boss! Will you be the Shadowblade (Developer), mastering the "vibe code" into precision strikes? Or perhaps the Summoner (Architect), using ADK and MCP to design a whole army of collaborating agents? You could be the Scholar (Data Engineer), wielding the powerful magic of RAG to fuel your creation with vast knowledge, or the Guardian (SRE), raising a shield of bulletproof AgentOps to ensure your agent survives the battle. Whatever your calling, a legendary adventure in building real-world AI awaits. Choose your path and join the fellowship! Begin Your Quest with the Agentverse Roadshow! After architecting the system, the Accelerate AI with Cloud Run program completes your journey by focusing on the Deploy and Scale stages. This developer day gives you hands-on experience taking your AI from prototype to production through three core labs. You will begin by deploying a secure MCP server on Cloud Run and building an agent with the ADK to communicate with it. Finally, you'll take the agent from prototype to production by containerizing and deploying it to a GPU-accelerated Cloud Run instance Sign Up for an Accelerated AI with cloud run Developer Days to build your agents for production and scale! More dates and locations in Europe, Middle East, Africa, and Asia-Pacific are coming soon. The End-to-End AI Platform in Action These programs are part of a bigger story: Google Cloud is an end-to-end AI platform. Here are recent resources showing our innovation at every stage: Get Hands-On with Your Data This month, we have some fantastic opportunities for developers looking to build innovative AI solutions on top of their data. Take the Challenge: Compete in the $100,000 BigQuery AI Hackathon to solve real-world problems with AI functions in your SQL environment. The Data Agent Journey: Start with the vision for a "Vibe Business Intelligence" data agent in this talk, then learn to build it in this deep-dive blog series. Core Platform and Dev Tools For Model Flexibility: The new Vertex AI Model Garden SDK makes using a vast library of open models simple. For Custom AI Workloads: Deploy a production-ready blueprint using the GKE inference reference architecture, then dive into our new AI Hypercomputer tutorials to serve Gemma 3 ,Llama 4, and Qwen3 on GKE or fine-tune Gemma 3 on Slurm or on GKE or train a model from scratch using Qwen2. Gemini CLI For Agent Memory: Learn about memory management in this post on Agent state and memory with ADK. For Hands-on Gemini CLI Practice: Follow the Hands-on with Gemini CLI codelab , learn how to create custom slash commands, this repo provides some great examples. One more thing Go Deeper: Your exclusive invitation to The Agent Factory 🎙️ Want to become an expert on AI agents? The Agent Factory is your guide. We started it to have the crucial conversations developers need about building production-ready AI agents, moving past the hype to discuss what really matters. We've already released episodes on: - Episode 1: Agents, Their Frameworks, and When to Use Them - Episode 2: Multi-Agent Systems: Concepts and Patterns - Episode 3: Building Custom Tools for Your AI Agent - Episode 4: Remember me: Memory in Agents The agent factory episode 3 - building custom tools for agents. Final Word The future belongs not to the AI scientist or the software engineer as separate roles, but to the 'AI builder' who can master the entire stack. The only question left is: What will you build? Follow us on Social media. Add your voice to the conversation and get the latest news, views, and insights. X Google Cloud blog LinkedIn Facebook YouTube GitHub No-cost cloud training 35 credits on Google Cloud Skills Boost. Redeem nowThe Overwhelmed Person’s guide to AI with Google Cloud: AugGoogle Cloud Innovators undefinedpfeifferj@archlinux.ch pfeifferj@archlinux.chZD