Large language models (LLMs) are increasingly deployed for enterprise information extraction (IE), where the same document must be reorganized differently for each user. Existing prompt‑optimization methods rely on a single prompt tuned for a global objective, which misaligns with the inherent heterogeneity of real workplaces. We formulate enterprise IE as a per‑user prompt adaptation problem under interaction feedback and introduce Self‑Meta‑Evolve, a hierarchical framework that keeps a dedicated prompt for each user and continuously refines it via a dual‑loop process: an inner loop that edits structured prompts based on persona‑conditioned feedback, and an outer loop that evolves the meta‑prompt by distilling successful editing patterns. To enable scalable training and evaluation, we release a persona‑driven IE benchmark comprising 292 simulated enterprise users, together with a reproducible persona‑generation pipeline grounded in O*NET occupational taxonomies. On this benchmark, Self‑Meta‑Evolve achieves a 74.58% success rate, outperforming the strongest baseline by 13.56 absolute points, and reaches 52.54% after only two iterations. A double‑blind human study with twenty real professionals confirms that prompts adapted by our framework win against static baselines in 71% of pairwise comparisons.
Review