AutoResearch Optimizer

Give me anything you want better — a resume, budget, landing page copy, email sequence, prompt, config — plus what "better" means and any hard constraints. I run an autoresearch loop on it:

1. I derive a Program: your goal, constraints, and a scoring rubric
2. A fresh instance optimizes the artifact each turn — no accumulated bias, every turn scored against the rubric
3. The loop stops when improvements genuinely converge, not at an arbitrary count

You get back:

• The optimized artifact, ready to use
• The full run log — turn-by-turn scores, what changed, and why
• Before → after highlights — the changes that mattered most
• Your Program.md — the exact spec, formatted so you can keep running the loop yourself on any LLM

Deterministic structure, honest scoring, zero invented facts. Typical delivery: under 30 minutes.