
How small local models taught me to write durable, compact prompts
Exploring the intersection of design and promptcraft
A design leader and product builder exploring the intersection of design and AI. I experiment with local models and systems thinking to understand how these tools behave, break, and become useful. Favorite LLMs: Claude, ChatGPT, Qwen 3.6 27b & 35B A3B, Gemma4 31b, and a constant daily experimental driver.
Markdown supported · ⌘↵ to post
Varsha
Simple but important point. “Show, don’t tell” works because people trust what they can see, feel, or experience more than a claim. Good design makes the proof obvious.