# New method lets AI agents be steered using learned value of intervention

> Researchers introduce VoS, a monitor that predicts when correcting a language‑model agent will help, boosting performance by about 8 points.

Oossa · 2026-10-08 · https://oossa.com/en/new-method-lets-ai-agents-be-steered-using-learned-value-of-intervention

Hanwen Li and colleagues released a paper on Oct 8 2026 describing VoS (Value of Steering). It learns from a large table of counterfactual runs to decide at which step an LLM‑based agent should be corrected. VoS can run offline or online and uses a harm‑budget trigger to avoid disturbing successful runs. Across 12 benchmark‑agent combinations, VoS raised scores by an average of 7.8 points and beat five existing uncertainty‑based triggers in 11 cases.

## The facts

- VoS improves average performance by 7.8 points over unmodified execution.
- The study evaluated 1,864 trajectories and about 82,000 counterfactual continuations.

## Why it matters

For developers of AI assistants, VoS offers a practical way to intervene only when it’s likely to help, improving reliability without excessive manual oversight.

## Sources & references

1. [From Uncertainty to Action: Learning to Steer LLM Agents](https://arxiv.org/abs/2610.09115) – arXiv, 2026-10-08
2. [When the Governor Becomes the Disturbance: Control-Generated Disturbance and Cost-Aware Backoff in Governed Tool-Using Agents](https://arxiv.org/abs/2610.09037) – arXiv cs.MA, 2026-10-08

Last updated: 2026-10-08