Step 1. Build “AI” (LLM models) that can’t be trusted, doesn’t learn, forgets instructions, and frustrates software engineers

Step 2. Automate the use of these LLMs into “agents”

Step 3. ???

Step 4. Profit