In this bonus episode of Talking AI, Omar Shanti, CTO of HatchWorks AI, breaks down how LangChain’s abstractions have both helped and hindered AI innovation.
He explains why these tools, while useful for quick starts, can sometimes make simple tasks harder as projects scale. The conversation highlights where LangChain shines and where it falls short.
Viewers will hear insights on why prompt engineering and tuning are more challenging than building agent tools. Omar shares how over-abstraction can cause issues when taking projects into production, leading many to rethink their toolkits.
If you’re working with LangChain or similar abstraction tools, this episode gives you practical advice on avoiding common pitfalls and understanding when these abstractions might not serve your needs.
- Introduction to LangChain’s abstractions
- What is abstraction?
- LangChain’s pros and cons
- Challenges with LangChain in production
- Observability and orchestration
- How much orchestration do you need?
- Final thoughts on using LangChain
