VeriTrace: Evolving Mental Models for Deep Research Agents
For developers of deep research agents, VeriTrace provides a novel explicit regulation mechanism that reduces error propagation and improves task understanding, outperforming implicit reasoning approaches.
VeriTrace introduces a cognitive-graph framework with three explicit regulatory loops (interpretive update, deviation feedback, schema revision) to regulate mental model evolution in deep research agents, achieving 4.22 pp improvement on DeepResearch Bench Insight and 5.9 pp overall win rate on DeepConsult over matched baselines.
Deep research agents face vast, interdependent, and pervasively uncertain information. Existing systems explore what evolving intermediate representations should look like, but leave their evolution to the LLM's implicit reasoning. Without explicit regulation, the intermediate layer is easily contaminated by mixed-quality information and propagates errors along its dependencies, so model scale often ends up substituting for absent regulation. We argue that an agent's mental model should instead evolve through explicit feedback that continuously aligns task understanding with reality, and identify three regulatory loops: interpretive update, deviation feedback, and schema revision. We realise this in VeriTrace, a cognitive-graph framework that explicitly implements the three loops. Using matched Qwen3.5-27B backbones, VeriTrace improves over the strongest matched baseline by 4.22 pp on DeepResearch Bench (DRB) Insight (1.49 pp Overall) and by 5.9 pp Overall win rate on DeepConsult. With Config-DeepSeek, it achieves the strongest reproducible open-source result on DRB.