# Agentic generation of verifiable rules for deterministic, self-expanding reaction classification

arXiv: https://arxiv.org/abs/2607.01061 · HTML: https://mlsanity.com/paper/2607.01061/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Medical and scientific domain agents, one of 16 directions in LLM agents and tool use.
- Builds on 19 of its 52 references that are in the map.
- First posted on arXiv 1 Jul 2026.

## Placed in

- LLM agents and tool use / [Medical and scientific domain agents](https://mlsanity.com/agents/medical-and-scientific-domain-agents/)

## Builds on (most cited first, 19 of 19)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [ChemCrow: Augmenting large-language models with chemistry tools](https://mlsanity.com/paper/2304.05376/) (2023)
- [Confident Learning: Estimating Uncertainty in Dataset Labels](https://mlsanity.com/paper/1911.00068/) (2019)
- [Chain-of-Verification Reduces Hallucination in Large Language Models](https://mlsanity.com/paper/2309.11495/) (2023)
- [Molecular Transformer - A Model for Uncertainty-Calibrated Chemical Reaction Prediction](https://mlsanity.com/paper/1811.02633/) (2018)
- [Learning to Plan Chemical Syntheses](https://mlsanity.com/paper/1708.04202/) (2017)
- [Context Length Alone Hurts LLM Performance Despite Perfect Retrieval](https://mlsanity.com/paper/2510.05381/) (2025)
- [Predicting Organic Reaction Outcomes with Weisfeiler-Lehman Network](https://mlsanity.com/paper/1709.04555/) (2017)
- [Mapping the Space of Chemical Reactions Using Attention-Based Neural Networks](https://mlsanity.com/paper/2012.06051/) (2020)
- ["Found in Translation": Predicting Outcomes of Complex Organic Chemistry Reactions using Neural Sequence-to-Sequence Models](https://mlsanity.com/paper/1711.04810/) (2017)
- [TaxoAdapt: Aligning LLM-Based Multidimensional Taxonomy Construction to Evolving Research Corpora](https://mlsanity.com/paper/2506.10737/) (2025)
- [Electron flow matching for generative reaction mechanism prediction obeying conservation laws](https://mlsanity.com/paper/2502.12979/) (2025)
- [Synthelite: Chemist-aligned and feasibility-aware synthesis planning with LLMs](https://mlsanity.com/paper/2512.16424/) (2025)
- [Holistic chemical evaluation reveals pitfalls in reaction prediction models](https://mlsanity.com/paper/2312.09004/) (2023)
- [Atom-anchored LLMs speak Chemistry: A Retrosynthesis Demonstration](https://mlsanity.com/paper/2510.16590/) (2025)
- [SynthStrategy: Extracting and Formalizing Latent Strategic Insights from LLMs in Organic Chemistry](https://mlsanity.com/paper/2512.01507/) (2025)

## Built on by (most cited first, 1 of 1)

- [Strategy-first synthesis planning for complex natural products](https://mlsanity.com/paper/2608.07454/) (2026)

## Abstract (arXiv)

Computer-assisted synthesis planning breaks target molecules into accessible precursors using large libraries of reaction rules that assign each transformation a deterministic, interpretable label. But chemistry is long-tailed, making manual encoding intractable, and existing tools rely on fixed rulesets that cannot adapt to new chemistries. Here we present a fully automated pipeline in which a multi-agent framework of large language models (LLMs) classifies reactions and writes the rules themselves across 665,901 US patent reactions, generating each rule under a verification loop that tests it against the corpus. It expands a standard taxonomy from 68 to 14,073 classes without human curation. With a lightweight fingerprint classifier, it classifies 97.7% of unseen reactions, matching a leading proprietary classifier while resolving chemistry more finely and extending on demand to chemistry outside its training distribution. The result is a living reactivity database and a general route to turning generative models into reliable, self-expanding symbolic systems.