Prototyping with Generative AI: Essential Patterns for Building Innovative Applications
Embark on a journey into the world of generative AI prototyping and discover the most impactful development patterns for building innovative applications using Google Cloud’s powerful tools. This session provides a practical guide to rapidly transforming your ideas into functional prototypes, empowering you to explore the potential of generative AI. In this session, here are the lessons.
- Rapid Prototyping with Gemini and Vertex AI: Quickly get started with generative AI development using the Gemini API and Vertex AI platform.
- The Art of Prompt Engineering: Learn prompt design and tuning techniques to elicit accurate and creative outputs from Gemini.
- Unlocking Multimodal Magic: Integrate various data types, including images and text, into your prototypes using Gemini’s multimodal capabilities.
- Building with Retrieval Augmented Generation (RAG): Leverage your existing data to enhance the knowledge and context of your prototypes with RAG.
Products Covered:
Vertex AI Studio, Gemini 1.5 Pro/Flash, Model Garden, Model Builder, Workspace
Project IDX in beta
Complete developer workspace in the browser.
Building a server in 24 hrs.
Rag Retrieval augmented generation.
Scaling Generative AI: Advanced Patterns & Production Techniques
- Advanced Gemini Capabilities: Unlock the power of Gemini 1.5 Pro/Flash, exploring features like enhanced reasoning, function calling and AI agents to build more sophisticated and interactive applications.
- Integrating Visuals with Imagen 3: Learn how to integrate Imagen 3 for advanced image generation and understanding, creating visually compelling and interactive user experiences within your applications.
- Building with Function Calling and AI Agents: Dive deeper into function calling and AI agents, exploring how to connect your application to external APIs and data sources, enabling dynamic interactions and automation.
- Optimizing for Production with Context Caching and Deployment Strategies: Discover techniques like context caching to enhance performance and efficiency, and explore best practices for deploying and scaling your generative AI applications on Google Cloud for production readiness.
Products Covered:
Vertex AI, Gemini 1.5 Pro/Flash, Imagen 3, LangChain, LangGraph, Agent Builder, Cloud Run
- Data
- Model
- Application
- Infrastructure
The new genre of computing equals a more random model
GENERATIVE AI
of computing. It is good to have a random output.
That way it isn’t too robotic.
With objective deterministic outputs and subjective probabilistic outputs, what problems can be uniquely solved by GenAI?
The IVO test
If you can immediately validate the output it is a good application.
Generative ai
Legacy code generator creates static code analysis to perform static architecture.
Decoding AI’s potential with Anthropic. A guide for technical business leaders.
Anthropic
Experts in engineering helped transform us to AI to function autonomously.
How to work with Claude
The upgrade to Claude 3.5 sonnet simulates a real work environment. With real world use cases we can look at a large scale code base.
The future impact of generative AI is yet to be seen. With proponents on both sides, do you think AI is necessary in order to advance technology?
Getting Started with Generative AI. ’Getting Started with Generative AI’, Zack Akil Sr. Full Stack Machine Learning Engineer & Developer Advocate. Google. (2024, November).
Welcome: Startup School. ’Welcome: Startup School’, Tymon Kokoszka, Go-To-Market Manager, Google for Startups. Google. (2024, November).
Deep dive: Vertex AI for ML practitioners. ’Deep dive: Vertex AI for ML practitioners’, Mona Mona, AI/ML Customer Engineer, Google Cloud. (2024, November).
Resources
Google Startup school https://cloudonair.withgoogle.com/events/startup-school
Vertex AI Studio https://cloud.google.com/generative-ai-studio
Vertex AI Platform https://cloud.google.com/vertex-ai
Generative AI Feedback goo.gle/GenAI-FeedbackW1
Gemini Google gemini.google.com