arXiv:2511.11519v1 Announce Type: cross 
Abstract: Enabling agentic AI systems to adapt their problem-solving approaches based on post-training interactions remains a fundamental challenge. While systems that update and maintain a memory at inference time have been proposed, existing designs only steer the system by modifying textual input to a language model or agent, which means that they cannot change sampling parameters, remove tools, modify system prompts, or switch between agentic and workflow paradigms. On the other hand, systems that adapt more flexibly require offline optimization and remain static once deployed. We present Experience-Guided Reasoner (EGuR), which generates tailored strategies -- complete computational procedures involving LLM calls, tools, sampling parameters, and control logic -- dynamically at inference time based on accumulated experience. We achieve this using an LLM-based meta-strategy -- a strategy that outputs strategies -- enabling adaptation of all strategy components (prompts, sampling parameters, tool configurations, and control logic). EGuR operates through two components: a Guide generates multiple candidate strategies conditioned on the current problem and structured memory of past experiences, while a Consolidator integrates execution feedback to improve future strategy generation. This produces complete, ready-to-run strategies optimized for each problem, which can be cached, retrieved, and executed as needed without wasting resources. Across five challenging benchmarks (AIME 2025, 3-SAT, and three Big Bench Extra Hard tasks), EGuR achieves up to 14% accuracy improvements over the strongest baselines while reducing computational costs by up to 111x, with both metrics improving as the system gains experience.

يتناول المقال نظام Experience-Guided Reasoner (EGuR)، وهو نظام ذكاء اصطناعي جديد مصمم لتكييف استراتيجيات حل المشكلات بناءً على التجارب المتراكمة خلال وقت الاستدلال. على عكس الأنظمة الحالية التي تعدل المدخلات النصية فقط، يقوم EGuR بإنشاء استراتيجيات مخصصة بشكل ديناميكي، مما يسمح بنهج أكثر مرونة في التفكير الاصطناعي. هذه الخطوة تعالج التحدي المتمثل في تمكين أنظمة الذكاء الاصطناعي الوكيلة من تكييف منهجياتها بعد التدريب.

El artículo discute el Experience-Guided Reasoner (EGuR), un nuevo sistema de IA diseñado para adaptar sus estrategias de resolución de problemas en función de las experiencias acumuladas durante el tiempo de inferencia. A diferencia de los sistemas existentes que solo modifican las entradas textuales, EGuR genera estrategias personalizadas de manera dinámica, permitiendo un enfoque más flexible para el razonamiento en IA. Este avance aborda el desafío de permitir que los sistemas de IA agentiva adapten sus metodologías después del entrenamiento.

L'article traite de l'Experience-Guided Reasoner (EGuR), un nouveau système d'IA conçu pour adapter ses stratégies de résolution de problèmes en fonction des expériences accumulées pendant le temps d'inférence. Contrairement aux systèmes existants qui ne modifient que les entrées textuelles, EGuR génère des stratégies sur mesure de manière dynamique, permettant une approche plus flexible du raisonnement en IA. Cette avancée répond au défi de permettre aux systèmes d'IA agentique d'adapter leurs méthodologies après l'entraînement.

The article discusses the Experience-Guided Reasoner (EGuR), a novel AI system designed to adapt its problem-solving strategies based on experiences accumulated during inference time. Unlike existing systems that only modify textual inputs, EGuR generates tailored strategies dynamically, allowing for a more flexible approach to AI reasoning. This advancement addresses the challenge of enabling agentic AI systems to adapt their methodologies post-training.

Experience-Guided Adaptation of Inference-Time Reasoning Strategies

Was this article worth reading? Share it

Ready to build your own newsroom?