Can a routing harness generate its own training data automatically?
Explores whether the logs and signals produced by an agent routing system—which directs requests to appropriate model tiers—naturally contain the evidence needed to improve the models themselves through fine-tuning and distillation.
The paper starts from the requirement that recursive self-improvement needs "a concrete mechanism through which a system observes its own capabilities and converts that evidence into the next round of learning." Its claim is that a deployed routing harness already is one: "beyond task outputs, agentic interaction leaves execution trajectories together with observable evidence of what a model can and cannot yet do." The discussion lists three reusable signals the harness produces while serving requests: execution trajectories that "ground training in real interaction," routing signals that "characterize capability demand," and recorded outcomes that "reveal where the model still falls short."
The mechanism as stated is a heterogeneous model pool behind intelligent routing. For every turn the system records the capability demand predicted, the service tier selected, and the interaction that followed. Those records become user-turn training examples that keep interleaved reasoning, tool calls, and harness context, and they are admitted only after structural validation, six-dimensional semantic evaluation, and subscene-level labeling. The routing signals then do a second job. They "provide estimates of capability demand," which the paper uses to organize supervised fine-tuning into a three-stage curriculum and to extend "naturally" to routing-guided on-policy distillation, where a teacher supervises student-generated responses "under the same staged progression." The router is therefore both a serving component and a difficulty labeler for the training data it helps produce.
This sits on the other side of the slow/fast split from the AIDE2 loop. The AIDE2 note places recursive self-improvement at the harness layer, where the agent rewrites its own code and the excerpt mentions no weight updates (How does an AI agent improve its own research code?). NeoHorse-1 keeps the harness as an instrument and sends what it observes into fine-tuning and distillation, which is the parametric loop in the survey's taxonomy (Do self-improving agents really split into two distinct loops?). SkillClaw starts from the same raw material, interaction trajectories, but converts them into skills synchronized back to agents (How can agent systems share learned skills across users?). My reading, not the paper's: the two-loop taxonomy describes what the conversion step outputs, not where the evidence comes from, since one harness log could feed either loop.
The excerpt does not establish that the loop is recursive. It gives no results, model sizes, or benchmarks, and the abstract frames the system as built "to explore this path." It does not say whether the trained model returns to the pool and shapes later routing and data, which is what would separate a compounding loop from a one-way data pipeline. It does not say who the teacher in the distillation setup is, or whether the routing policy is itself updated. The introduction also notes that trajectories and executable tasks "have already been used to train agentic models," so what the paper adds is the routing signal as a capability-demand label, not trajectory logging. Read at that strength, the note records a design claim: routing metadata gives a deployed system a cheap, per-turn estimate of demand that can order training data. Whether that yields self-improvement is left open by the excerpt.
Inquiring lines that read this note 27
This note is a source for these research framings, grouped by the broader line of inquiry each explores. Scan the bold lines of inquiry; follow any specific question forward.
Can intelligent model routing compete with model scaling as an optimization lever?- Does model selection matter more than model improvement for query routing?
- How do cost-aware cascades compare to single-turn routing in the component framework?
- Can routing signals organize training data into a meaningful curriculum automatically?
- Does curating smaller model pools improve router recall compared to scaling?
- Can routing ever beat model improvement as a performance lever?
- Do engineering costs of routing layers offset the savings from cheaper models?
- What learning signals best supervise router training across benchmark tasks?
- Who serves as the teacher model in the routing-guided distillation process?
- Does the improved model actually return to the routing pool and shape future decisions?
- Why does a second routing level sometimes break accuracy in disclosure hierarchies?
- How much worse are convergent routing methods than theoretical oracle routing?
- Can clustering-based routing match neural network routers in cost and quality?
- Can runtime behavior mapping help localize harness deficiencies?
- What makes a harness low-friction for model strategy?
- How can decision quality be automatically extracted from agent trajectories?
- What makes next-state signals from agent trajectories a reliable learning source?
- Can trajectory consistency predict real-world deployment failures?
Related concepts in this collection 3
This note in its neighbourhood — explore the map, then jump to a related concept in the list below.
Click a node to walk · click center to open · click Open in graph to see this note in the full knowledge graph
-
How does an AI agent improve its own research code?
Explores the feedback loop where an AI research agent modifies and tests its own codebase, with each successful change becoming the agent that proposes the next revision. This specificity matters because it distinguishes a narrow, defined mechanism from broader claims about open-ended self-improvement.
contrast: AIDE2 rewrites harness code, while NeoHorse-1 uses the harness as a source of training evidence
-
Do self-improving agents really split into two distinct loops?
Explores whether modern self-improving agents can be understood through a clean abstraction separating fast scaffold updates from slow model weight updates, and whether this framework actually explains the field's recent progress.
places this design on the parametric side, fed by scaffold-side observation
-
How can agent systems share learned skills across users?
Individual users operating autonomous agents independently rediscover solutions because systems lack mechanisms to propagate discoveries. Can centralized aggregation and automatic evolution convert isolated experiences into shared capabilities?
same raw material, interaction trajectories, converted into skills instead of training examples
Related papers in this collection 8
Papers most semantically related to this note, ranked by cosine similarity in the embedding space.
- NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness
- SoL-Pi: Recursively Scaling Auto-Research Loops for Efficient Agent Harness
- StateM: Reaching 95.3% Raw Accuracy, or a \$15 Frontier Run, on Terminal-Bench 2.1 via Harness Scaling
- Harness Handbook: Making Evolving Agent Harnesses Readable,Navigable, and Editable
- LLMRouterBench: A Massive Benchmark and Unified Framework for LLM Routing
- RSIGym: A Flexible Environment for Recursive Self-Improvement
- Scaling Laws for Agent Harnesses via Effective Feedback Compute
- RRSI: Regularized Recursive Self-Improvement of Agent Harnesses
Original note title
a deployed routing harness already contains the mechanism recursive self-improvement needs — trajectories, routing signals and outcomes become training data