AI EXPERIMENT

A learning file for testing how AI can support analysis workflows without replacing the thinking: prompt structure, small prototypes, review loops, and agentic coding practice.

AI tools can produce fast output, but speed is not the same as reliability. The challenge is designing workflows where AI helps explore, draft, and automate while the human still checks assumptions.

I treat each experiment like a small system: define the task, capture the prompt, inspect the output, test the result, and write down what failed. This keeps the work grounded instead of turning it into random prompting.

• Use AI for scaffolding and comparison, not blind final answers.
• Keep prompts and outputs close to the project so the process can be reviewed.
• Test generated code or analysis before trusting it.

A collection of small AI-assisted workflows, prototype notes, and reusable patterns for analysis and portfolio development.

AI becomes more useful when the workflow has friction in the right places: clear inputs, explicit checks, and a habit of asking what could be wrong.

Build a small demo that compares manual analysis steps with AI-assisted steps and documents where AI helped or got in the way.