Write the Finish Line

In the last issue I told you that an AI plan written more than 6 months ago is out of date. Because the models keep changing and improving, the advice from the Claude Code team itself was to delete your old instructions and rebuild from scratch. Now we're going to dive into how you might consider rebuilding.

Tracking LLM systems over time, you see interaction started with basic question-and-answer chatbots. Then it advanced to Projects for persistent context and Skills for consistent action. That was followed by the beginning of Agentic AI which added the ability to schedule processes and connect to Tools or external systems. Each step in this evolution has added more capabilities, and has given more autonomy to the AI system. And in each step, humans were responsible for writing and reviewing the details of the process to meet the objectives.

Now we have the option of letting AI decide how to meet the objectives. Instead of spelling out every step of the process to meet the objective, simply give AI the goal and let it figure it out and run on its own. This issue dives more into this newest advancement of Agentic AI.