.NET Rocks! Architectural Intelligence with Thomas Betts
Jan 9, 2025
Thomas Betts, a laureate software architect at Blackbaud and contributor to InfoQ.com, discusses the practical realities of integrating AI into software applications. He emphasizes that AI is essentially just another API, urging listeners to move beyond the hype. The conversation covers the potential of large language models as UX tools and the importance of precise communication in AI development. Betts also tackles the ethical concerns surrounding agentic AI, highlighting the need for oversight to ensure content accuracy.
AI Snips
Chapters
Transcript
Episode notes
Appropriate Use of LLMs
- Consider whether an LLM is appropriate for your software and how to optimize it if it is.
- LLMs excel in language-based tasks, such as natural language interfaces or simplifying complex UI.
Simplified Reporting UI
- A hackathon team at Thomas Betts's company used LLMs to simplify their custom report builder UI.
- The LLM translated natural language requests into API calls, improving user experience.
Checks and Balances for LLMs
- When using LLMs, implement checks and balances, like human review or auditing.
- Don't assume LLMs can handle everything autonomously, especially critical workflows.