mlsanityData to 14 Oct 2026

How to Train a Real-World Silicon Concierge? Internalizing Complex Business Workflow to Only OneModel

arXiv 2608.20350 ↗First posted 16 Jun 2026cs.CL cs.AIcited by 2

Not on the map. Why?

Abstract (arXiv)

Traditional industrial agents rely on modular pipelines, including Router, Retriever, Planner, Executor, Responder, Reviewer, and other components. These systems often fracture into a labyrinth of ad-hoc patches, leading to cascading errors and high latency. We propose OneModel, an applicable paradigm shift from external workflows to internalized knowledge representation. Unlike modular systems that slice fluid user intents into static steps, OneModel consolidates complex business logic and SOPs directly into the model parameters. Through Continual Pre-training (CPT) and logic-compilation SFT, we transform fragmented business rules into intuitive model reasoning within a unified attention space. Deployed in our global financial service system, OneModel effectively breaks the trade-off between latency, accuracy, and complexity. Online A/B testing demonstrates an end-to-end latency reduction of more than 50 percent, from 18.7 seconds to 8.0 seconds, while the Intelligent Resolution Rate (IRR) increases from 64.3 percent to 83.3 percent. The results show that OneModel can replace brittle engineering logic with internalized cognitive intuition, offering a scalable blueprint for transitioning industrial agents from complex, error-prone workflows to unified model architectures.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 29 of its 39 references that are in the map. First posted on arXiv 16 Jun 2026.

Builds on

29 of 39 references.

References · 29

  1. 2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
  2. 2023Efficient Memory Management for Large Language Model Serving with PagedAttentioncited by 5,944
  3. 2023Let's Verify Step by Stepcited by 3,789
  4. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,455
  5. 2020Don't Stop Pretraining: Adapt Language Models to Domains and Taskscited by 1,565
  6. 2022Measuring and Narrowing the Compositionality Gap in Language Modelscited by 1,075
  7. 2023Large Language Models Cannot Self-Correct Reasoning Yetcited by 1,017
  8. 2022Solving math word problems with process- and outcome-based feedbackcited by 901
  9. 2023Enhancing Chat Language Models by Scaling High-quality Instructional Conversationscited by 839
  10. 2023DSPy: Compiling Declarative Language Model Calls into Self-Improving Pipelinescited by 832
  11. 2025SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-trainingcited by 672
  12. 2025τ2-Bench: Evaluating Conversational Agents in a Dual-Control Environmentcited by 627
17 more
  1. 2022Large Language Models Struggle to Learn Long-Tail Knowledgecited by 458
  2. 2023Making Retrieval-Augmented Language Models Robust to Irrelevant Contextcited by 369
  3. 2023Character-LLM: A Trainable Agent for Role-Playingcited by 339
  4. 2025Demystifying Long Chain-of-Thought Reasoning in LLMscited by 326
  5. 2023ToxicChat: Unveiling Hidden Challenges of Toxicity Detection in Real-World User-AI Conversationcited by 231
  6. 2021Mind the Gap: Assessing Temporal Generalization in Neural Language Modelscited by 230
  7. 2024Can LLMs Learn from Previous Mistakes? Investigating LLMs' Errors to Boost for Reasoningcited by 69
  8. 2025Know You First and Be You Better: Modeling Human-Like User Simulators via Implicit Profilescited by 41
  9. 2023PlatoLM: Teaching LLMs in Multi-Round Dialogue via a User Simulatorcited by 29
  10. 2024DuetSim: Building User Simulator with Dual Large Language Models for Task-Oriented Dialoguescited by 24
  11. 2021Domain-independent User Simulation with Transformers for Task-oriented Dialogue Systemscited by 23
  12. 2024BPO: Towards Balanced Preference Optimization between Knowledge Breadth and Depth in Alignmentcited by 20
  13. 2023Eliminating Reasoning via Inferring with Planning: A New Framework to Guide LLMs' Non-linear Thinkingcited by 19
  14. 2025Understanding and Mitigating Overrefusal in LLMs from an Unveiling Perspective of Safety Decision Boundarycited by 18
  15. 2024Optimizing Language Model's Reasoning Abilities with Weak Supervisioncited by 15
  16. 2025Rethinking Reasoning Quality in Large Language Models through Enhanced Chain-of-Thought via RLcited by 5
  17. 2025Dingtalk DeepResearch: A Unified Multi Agent Framework for Adaptive Intelligence in Enterprise Environmentscited by 5

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026ARC: Fair Relative Advantage Comparison in Open-Ended Real-World Interactioncited by 0
  2. 2026Grounded Skill Synthesis from Code at Scale for Agentic Intelligencecited by 0