Agentic AI
Systems that plan, delegate, use tools, and operate beyond single-turn chat.
The essays collected here treat agentic AI less as a mysterious property of a model than as a new software form. Claude Cowork makes the shift visible because the system does not merely answer prompts; it plans, executes, and loops through work. The Fork in the Road: Claude Code vs CoWork sharpens the point by showing that architecture, permissions, and resumability shape what an agent can become just as much as raw intelligence does.
Read alongside Prompt, Context, Harness, the larger argument is that agents force software to be rethought around execution and coordination. The important question is no longer whether an AI can produce a strong answer in one turn, but whether a system can stay coherent across tools, sub-agents, permissions, and changing state.
Related
The layer of prompts, context, tools, and governance that makes multi-step AI work reliable.
The emerging ecology of terminal agents, skills, interfaces, and machine-native software tooling.
A path through historical voice machines, modern TTS, and the social uses of synthetic speech.
Read Next
- Claude Cowork
Claude Cowork marks Anthropic's pivot from chatbot to agentic AI — a productized Claude Code that plans, executes, and delegates tasks through sub-agents in a sandboxed Micro-VM environment.
- The Fork in the Road: Claude Code vs CoWork
Two products born from the same codebase diverge in fundamental ways. Claude Code's local execution and CoWork's managed VM architecture shape not just performance and cost, but determine three critical capabilities—resumability, attachability, and autonomy—that reveal what each system can ultimately become.
- Prompt, Context, Harness: The Three Phases of AI Engineering
AI engineering has evolved through three compensatory phases — prompt, context, and harness — each addressing a failure the previous layer couldn't fix. Harness Engineering is the governance layer that keeps teams of agents coherent across complex, long-running tasks.
