LLMs as the new high level language
Summary
Explores the idea that LLM agents could become the new high-level programming language, enabling humans to build much more with autonomous agents. It discusses design patterns (documentation, implementation, dialogs, tasks), the MCP tool-calling standard, and the shift toward a documentation-centric, agent-driven software stack that breaks siloed apps. The post weighs objections about quality and understandability and emphasizes the need for reliable, auditable outputs.