Tag
The article synthesizes recent data and studies on LLMs in software development, revealing significant limitations in autonomy, context handling, and real-world outcomes, suggesting AI coding amplifies existing strengths rather than fixing weaknesses.
Cognition launches an official ambassador program for DevinAI, inviting applications for 50 new global ambassadors who receive free Devin Max plans, credits, and early access.
An essay exploring how LLMs have shifted the buy-vs-build calculus for software, arguing that while AI lowers build costs, maintenance and human oversight still make building a non-trivial investment, using the example of replacing Jira with an LLM-built internal tool.
Andrej Karpathy discusses how AI-generated software is increasing demand for software, referencing Jevon's paradox.
Andrew Ng highlights how AI is shifting software development from coding to clear communication through prompts. The author shares their experience building a web app with AI tools like VS Code, Kombai, and GitHub.
OpenAI shares how its team built a full software product with zero manually-written code using Codex agents, focusing on designing environments and feedback loops for reliable agent operation.