Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

Agentic Routing: The Harness-Native Data Flywheel

Source

arXiv

arxiv.org/abs/2607.11399v1

Read the source

Source

  • Source name: arXiv
  • Source kind: paper
  • Authors / organization: Xinchen Liu, Hang Zhou, Yingjie Zong, Yuchuan Tian, Liuyang Song, Shuo Zhang, Yulong Li, Wei He, Mengyu Zheng, Runke Liu, Siyang Cheng, Xiang Kuang, Hailin Hu, Kai Han, Yunhe Wang
  • Venue: arXiv: cs.CL
  • Published: 2026-07-13
  • Collected: 2026-07-14
  • URL: http://arxiv.org/abs/2607.11399v1
  • PDF: https://arxiv.org/pdf/2607.11399v1
  • DOI: UNKNOWN

Plain-Language Brief

The authors propose a new method for routing large language model agents called Harness-Native agentic routing. This method selects either a single best-fit model or multiple complementary models based on the full harness state, which includes observations, context, control, actions, and states. The goal is to improve cost-quality trade-offs and create a 'data flywheel' for better execution traces that can train better routers and models.

What the Document Says

  • The paper discusses the execution of large language model agents through an execution harness.
  • It mentions that these agents are becoming structurally specialized, meaning they excel in specific areas but not necessarily all.
  • Model selection inside an agent is considered a core systems problem.
  • Existing routing methods do not consider the execution state, intermediate failures, and feedback loops that make agents different from chat completion.
  • The authors propose Harness-Native agentic routing, which selects models based on the full harness state.
  • Every routing decision produces a structured data record for training better routers and harness-native models.
  • These records form a 'data flywheel' that improves cost-quality trade-offs and generates more traces under the same budget.
  • The authors instantiate this idea in OpenSquilla with a four-layer routing stack, an open LightGBM cold-start ranker, and a staged router-model path.

Key Concepts Mentioned

  • large language model agents
  • execution harness
  • cost-quality trade-offs
  • Harness-Native agentic routing
  • data flywheel

Claims Made by the Document

  • Existing routing methods do not consider the execution state, intermediate failures, and feedback loops that make agents different from chat completion.
  • The proposed Harness-Native agentic routing selects either a single best-fit model or multiple complementary models based on the full harness state.

Evidence Used

  • Existing methods optimize single-turn cost-quality trade-offs and therefore miss the execution state, intermediate failures, and feedback loops that make agents different from chat completion.
  • The proposed method naturally produces a structured data record consisting of the query, harness state, model choice or model set, execution trace, outcome, and cost.

Methods / Approach

  • Harness-Native agentic routing
  • four-layer routing stack
  • open LightGBM cold-start ranker
  • staged router-model path

Limitations Stated by the Authors

  • The authors argue that agentic routing is not merely cost control, but a data engine for agent-native training.

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Large language model agents are increasingly executed not by a single model call, but by an execution harness that manages observation, context, control, action, state, and verification. At the same time, frontier and open models are becoming structurally specialized: a model that is strong at code editing, long-context recovery, tool use, mathematical reasoning, or low-latency response may not dominate on the other axes. This makes model selection inside an agent a core systems problem rather than a per-query serving trick. Existing routing methods mostly optimize single-turn cost-quality trade-offs and therefore miss the execution state, intermediate failures, and feedback loops that make agents different from chat completion. We propose Harness-Native agentic routing, a step-level routing paradigm that selects either a single best-fit model for cost-effective execution or multiple complementary models for ensemble-style accuracy improvement, conditioned on the full harness state. The key insight is that every routing decision naturally produces a structured data record -- consisting of the query, harness state, model choice or model set, execution trace, outcome, and cost -- whose labels are supplied by the environment rather than by the router itself. These records form a harness-native data flywheel: execution traces train better routers and harness-native models, which improve cost-quality trade-offs and generate more traces under the same budget. We instantiate this idea in OpenSquilla with a four-layer routing stack, an open LightGBM cold-start ranker, and a staged router-model path that turns logged arena records into progressively stronger routing policies. The report studies singleton and multi-model routing on agentic benchmarks including DRACO and PinchBench, and argues that agentic routing is not merely cost control, but a data engine for agent-native training.