Domain expertise has always been the real moat (brethorsting.com)

Article summary

The article argues that domain expertise has always been the key factor in writing software, as it requires a deep understanding of the domain to build a working model. With the advent of agentic AI, the link between domain expertise and coding has been severed, allowing non-coders to produce software. However, the article claims that domain expertise is still essential to verify the correctness of the software. The most valuable person in this new world is the one who has both coding skills and domain expertise.

Main themes

• Domain expertise
• Agentic AI
• Software development
• Coding skills
• Automation
• Job displacement

What commenters say

• Domain expertise is still essential to verify the correctness of software, even with the help of agentic AI.
• Agentic AI can encode domain knowledge, allowing non-experts to become proficient in a domain quickly.
• Some commenters disagree, arguing that becoming a domain expert requires more than just querying an LLM.
• The value of domain expertise lies in its ability to provide ground truth, which AI systems lack.
• The rise of agentic AI will lead to job displacement, but some fields may be more resistant to automation than others.
• The importance of domain expertise is not limited to coding, but also applies to other fields such as civil engineering and e-commerce.
• Some commenters argue that the article overestimates the impact of agentic AI on the job market, and that other factors such as economic recession are more significant.
• The ability to write software is not the same as having domain expertise, and the two skills are distinct and valuable in their own right.