Analog IC layout is one of the hardest AI benchmarks: spatial reasoning, multi-objective tradeoffs (matching, parasitics, routing), no automated P&R tools. We ran VizPy's ContraPrompt on it. The optimizer mines failure→success pairs across iterations, extracting layout strategy rules the LLM learns to apply. Result: 97% of expert placement quality. Outperforms RL fine-tuning of a 120B model by 26%. No domain-specific training data. vizops.ai/blog/prompt-optimi…

We gave an LLM a 9-line random-search stub and a blackbox objective. 5 rounds of contrastive feedback later, it writes a solver that beats Optuna on 96% of benchmarks (53/55 EvalSet problems) at the same 2k eval budget. The key: contrastive pairs surface landscape structure raw scores don't. The LLM learns geometry from paired failures vs successes — then rewrites its strategy each round. Final code: 9 lines → 100-230 lines of specialized multi-phase optimization. No hand-tuning. vizops.ai/blog/contraprompt-…

Analog circuit placement is notoriously hard for LLMs. The constraints are spatial, parasitic-aware, and heavily design-rule-dependent. Even experienced engineers spend hours per block. We tried prompt optimization on it. Results surprised us.

Standard prompting gives you generic layout heuristics. VizPy's ContraPrompt mines failure→success pairs from actual placement attempts — extracting what matters: symmetry constraints, proximity rules, shielding logic. The LLM starts reasoning like a layout engineer instead of a chatbot.