Post by @sarainwondertech

Claude Code is only 1.6% AI. So what is the other 98.4%? When Anthropic accidentally leaked its own source code, researchers went through the whole thing and found that only around 1.6% of it is the actual AI decision-making. Most people think the model is where all the AI magic happens. But the model is only one piece of the system. The real product is everything wrapped around it: tools, permissions, memory, context, checks, workflows, and all the boring software engineering that turns raw AI output into something useful. Without that layer, you don’t have an agent. You have a chatbot. That’s the part most people miss. And it’s also why so many AI projects fail: companies buy access to a model and expect magic, but never build the system around it. If this news to you, chances are it is news for most people, so share it and hit follow for more well researched AI content. Paper: Dive into Claude Code: The Design Space of Today’s and Future AI Agent Systems Authors: Jiacheng Liu, Xiaohan Zhao, Xinyi Shang, and Zhiqiang Shen. #airesearch #claudecode #harness #aiagents