Course
Digicomp Code AGTSPC
Effective Prompting & Google AgentSpace and NotebookLM («AGTSPC»)
Learn the fundamentals of prompting LLMs and exploring further techniques for improving their output. Google Agentspace and NotebookLM Enterprise are designed to enhance information discovery, knowledge management, and productivity.
Duration
1 day
Price
850.–
Course documents
Official Google Cloud courseware
Course facts
- Understanding the basics of generative AI and how it compares to traditional AI
- Designing effective prompts following general best practices
- Improving LLM outputs using easy-to-approach prompt engineering techniques
- Writing effective prompts for interacting with multimodal models such as Gemini Vision Pro
- Describing the value of Google Agentspace and NotebookLM Enterprise to an organization
- Using Agentspace to find documents, get answers to questions, brainstorm, and perform actions
- Explaining NotebookLM's main features and its differences from NotebookLM Enterprise
- Creating an Agentspace app and configuring an identity provider
- Granting access to Workspace data stores and discussing use cases for NotebookLM Enterprise
- Define generative AI as a subset of AI
- Discuss the potential impact of AI on organizations
- Provide examples of generative AI applications
- Understand the role of context and examples in prompts
- Iterate on prompts to improve model output
- Describe the difference between prompt design and prompt engineering
- Write prompts using few-shot and chain-of-thought techniques
- Learn about additional prompt engineering techniques being used today
- Understand the multimodal models and their use cases
- Design effective prompts for multimodal models
- The problem of information discovery in enterprises
- How Agentspace extracts insights from enterprise data
- Agentspace as a homepage for daily work
- Searching for documents and content
- Generative AI summaries
- Question answering
- AI assistance for guidance
- Actions in connected systems
- Announcements
- Custom agents
- Retrieval-Augmented Generation (RAG)
- Vector Search and embeddings
- Semantic relationships in embeddings
- Agentspace's embedding process
- Summary of grounded responses
- Agentspace apps on Vertex AI
- Agentspace as a no-code solution
- Data stores and data source access control
- Data store types, formats, and sources
- Identity provider integration
- App features for search and assistance
- Agentspace homepage preview
- Customizing the user interface
- Assistant behavior configuration
- Homepage access with integrations
- Usage insights with Analytics
- Actions to connect to external services
- First-party and third-party action types
- OAuth consent screen and client setup
- The challenge of going from information to insight
- NotebookLM as a virtual research assistant
- How NotebookLM enhances research: summarization, question answering, idea generation
- Use case examples
- Main screen overview: Sources, Chat, Studio
- Adding and selecting sources
- Source summary and key topics
- Chat interactions and citations
- Saving responses as notes
- Creating and managing notes
- Studio features: Audio Overview, Briefing doc, FAQ, Timeline, Deep Dive conversation
- NotebookLM editions: NotebookLM, NotebookLM Enterprise, NotebookLM Pro
- Enterprise-grade security and privacy
- Comparison of usage limits
- Advanced chat settings: Conversation style and response length
- Notebook analytics
- Sharing a "Chat only" notebook
- NotebookLM for various functions: Sales, Training, Teaching, Customer Support, Marketing, Development
- Examples of notebooks for specific use cases
- Streamlining onboarding and team collaboration
- Creating pre-defined outputs
- Customizing chats
- Audio Overviews for learning
- Checklist and considerations: Purpose, collaboration, sources
- Upload process for different source types
- Re-syncing Google Drive files
- Handling websites and text-based files
- Managing YouTube videos and audio files
- Tips for getting started: Connecting materials, customizing audio, summarizing chats