Tech Talks with Kimberly and AIJoe

The conversation delves into the design considerations for embedding models, storage and management of embedding models, model versioning, separate tables for embedding models, switching data and model cut-over, strategic partitioning for data analysis, handling chunking and provenance tracking, recency and legal requirements, read, write, and data structure considerations, and policy-based management for agents.

Takeaways

• Embedding models require careful design and consideration
• The use of separate tables for embedding models allows for flexibility and better management

Chapters

00:00 Exploring Vector Use Cases

02:44 Embedding Storage Strategies

05:29 Managing Embedding Models

08:10 Data Partitioning and Embedding

10:42 Designing for Change in Embeddings

13:36 Best Practices for Database Agents

The conversation delves into the use of vector search for identifying schema and table structure similarities, highlighting its benefits and limitations. It explores the challenges of using vector search to identify redundancy and foreign key relationships within databases.

Takeaways

• Vector search for schema and table structure similarities
• Limitations of vector search in identifying redundancy and foreign key relationships

Chapters

• 00:00 Schema and Table Structure Similarities

The conversation explores the power of AI and vector search, focusing on code exploration and analysis. It delves into the challenges of code sprawl, semantic similarity, and the potential of vector search to identify and address inconsistencies in code. The use of vector search as a tool for code analysis and refactoring is highlighted, along with its potential to enhance the role of DBAs and developers. The conversation also touches on the future role of AI agents in code search and the practical applications of vector search in database management.

Takeaways

• Code sprawl and semantic similarity pose challenges in code exploration and analysis.
• Vector search can be used to identify inconsistencies in code and facilitate refactoring.
• The use of vector search enhances the role of DBAs and developers in code analysis and refactoring.

Chapters

• 00:00 Introduction to Vector Search and Its Impact
• 02:59 The Role of DBAs in Vector Search
• 06:01 Agentic Coding and Its Implications
• 08:59 Use Cases for Vector Search in E-commerce
• 12:11 Prototyping with SQL Server and AI
• 14:59 The Future of Vector Search and Its Adoption
• 17:59 Conclusion and Final Thoughts

Why isn't Vector Search taking off as fast as we expected? We both agree that the power and complexity of vector search are both pros and cons - setting its adoption rate at a slower pace. However, we both feel that those who embrace this technology -- and especially those that master it -- will be able to move forward much faster than those who do not. Agentic coding, prototyping, and AI integration are really ramping up; those who leverage this technology quickly, effectively, and most importantly, correctly will see incredible benefits in productivity and ability to adopt exciting and powerful features / capabilities.

Unlike many other features that seemingly never went anywhere, vector search is here to stay. It's time to get motivated / educated in understanding and implementing vector search effectively.

Takeaways

• Vector search is a powerful feature with potential for data professionals
• The complexity of vector search requires education and understanding for effective implementation

Chapters

• 00:00 Introduction to Vector Search
• 06:01 Agentic Coding and Decision Making
• 13:02 The Impact on Database Administrators
• 18:59 Complexity and Potential of Vector Search

Kimberly and Joe dive into the capabilities of Claude Code, demonstrating its use for automation, natural language interaction, and custom skill development. It explores the integration, extensibility, and learning potential of Claude Code, emphasizing optimization, efficiency, and the challenges associated with its use. The conversation explores the potential of AI in various aspects of database management, business exploration, and creative endeavors. They also discuss the challenges and opportunities presented by AI-generated content and the role of consultants and researchers in leveraging AI effectively. The discussion highlights the need for understanding and utilizing AI tools to make day-to-day tasks more efficient and productive.

Takeaways

• Claude Code enables natural language interaction for task automation
• Custom skills and automation with Claude Code offer rapid prototyping and experimentation
• Claude Code provides opportunities for optimization, efficiency, and integration with external tools Exploring new business ideas with AI and Claude Code
• Leveraging AI for database management and creative endeavors

Chapters

• 00:00 Introduction to Claude Code
• 05:58 Custom Skills and Automation
• 12:09 Integration and Extensibility
• 18:06 Learning and Iteration with Claude Code
• 23:58 Optimization and Efficiency with Claude Code
• 30:05 Challenges and Considerations
• 36:16 Implementing Partitioning and Data Strategy
• 42:36 Enforcing Best Practices in AI-Generated Databases
• 47:46 The Role of Consultants and Researchers
• 53:08 AI Applications in Reading and Learning
• 59:46 Leveraging AI for Database Management

In this conversation, Joe and Kimberly delve into a discussion about the power and pitfalls of AI tools, exploring the challenges of AI reliability and the potential for AI tools to enhance productivity. The conversation covers the exploration of various AI tools and their applications, including Google AI Studio, Notebook LM, Cloud Code, OpenRouter, and more. The discussion delves into the use of temperature settings, document analysis, academic papers, storytelling, translation, language models, and API calls. Additionally, the conversation addresses the importance of trusting AI models and tools.

Takeaways

• AI tools for productivity
• Challenges of AI reliability AI Studio allows for the configuration of AI models using temperature settings.
• Notebook LM is a free resource for studying and working with language models.
• OpenRouter simplifies the use of API calls for different models and provides access to free open-source models.

Chapters

• 00:00 Reconnecting and Reflecting on the Past
• 31:04 Introduction to AI Studio and Temperature Settings
• 36:27 Use Cases of Notebook LM for Academic Papers and Research
• 44:16 Discussion on Mid Journey and Nano Banana AI Tools
• 53:31 Use Cases of OpenRouter for Model Selection and API Calls

The conversation covers a range of topics including exciting ventures and inspirations, travel adventures and wildlife encounters, technology and navigation, photography and conservation, and collecting / obsessions. The takeaways include discussions on citizen science and technology, travel experiences and photography, and food choices and dietary preferences. The conversation delves into the journey of collecting and letting go, exploring the emotional and psychological aspects of collecting items and the process of decluttering. It also touches on the impact of technology on learning and entertainment, highlighting the accessibility of educational content and the role of familiar entertainment in regulating emotions.

Takeaways

• Citizen science and technology
• Travel experiences and photography
• Food choices and dietary preferences The journey of collecting and letting go
• The impact of technology on learning and entertainment

Chapters

• 00:00 Exciting Ventures and Inspirations
• 10:12 Technology and Navigation
• 20:02 Photography and Conservation
• 27:06 Collecting and Obsessions
• 32:09 The Journey of Collecting and Letting Go

The conversation delves into the new AI capabilities introduced in SQL Server 2025, focusing on vector embeddings and semantic search. It explores the implementation, use cases, considerations, and security aspects of vector embeddings in SQL Server 2025.

Takeaways

• SQL Server 2025 introduces new AI capabilities
• Vector embeddings and semantic search are key features of SQL Server 2025

Chapters

• 00:00 Introduction to SQL Server 2025 AI
• 06:12 Use Cases and Considerations for Vector Embeddings
• 19:29 Security and Compliance Considerations

The conversation delves into the concept of vector embedding models, their applications, integration with SQL Server, use cases, and associated challenges and considerations. It explores the power of vector embedding models for semantic search, fuzzy search, anomaly detection, and contradiction detection, providing valuable insights into the evolving landscape of data management and search capabilities.

Takeaways

• Vector embedding models enable semantic search and comparison of data based on learned patterns and embeddings.
• Vector embedding models provide powerful capabilities for fuzzy search, anomaly detection, and contradiction detection.

Chapters

• 00:00 Understanding Vector Embedding
• 06:23 Integration with SQL Server
• 13:09 Hybrid Search and Use Cases
• 20:15 Challenges and Considerations

The conversation covers the frustration of photography, skepticism about AI, AI as a superpower, AI in storytelling and research, AI and information verification, AI and database integration, AI and vector search, AI and database integration challenges, the role of DBAs in AI integration, AI and search technologies, and AI and anomaly detection.

Takeaways

• AI can be a superpower for creative professionals
• DBAs play a crucial role in AI integration and database management

Chapters

• 00:00 The Frustration of Photography
• 06:26 AI and Database Integration
• 11:41 The Role of DBAs in AI Integration
• 18:05 AI and Anomaly Detection