Charts as Epistemic Actions for Experience-Driven Agentic Data Analysis
arXiv cs.AI · 22 Oct 2026, 03:26 AEST
Data agents are evolving to incorporate visual evidence in their analytical workflows, yet they still rely heavily on text for reasoning. Research shows that while agents can generate charts, they do not consistently utilize them effectively in their analysis.
Key facts
- Data agents built on ReAct-style loops now execute complex analytical workflows over structured data, yet their intermediate reasoning still runs almost entirely through text.
- We ask whether charts can serve as internal epistemic actions inside a data agent's reasoning loop, and how to teach an agent to use them.
- Agents score as well from self-generated charts as from numeric output, yet plotting access alone does not translate into consistent use.
- Task slices show that charts lead on temporal, pattern, and anomaly reasoning, while text leads on precise reading and conditional verification.
- On three held-out benchmarks with four LLM backbones, method{} raises analytical accuracy and visual evidence quality over matched baselines.
Read aloud
Data agents are evolving to incorporate visual evidence in their analytical workflows, yet they still rely heavily on text for reasoning. Research shows that while agents can generate charts, they do not consistently utilize them effectively in their analysis. Data agents built on ReAct-style loops now execute complex analytical workflows over structured data, yet their intermediate reasoning still runs almost entirely through text. We ask whether charts can serve as internal epistemic actions inside a data agent's reasoning loop, and how to teach an agent to use them. Agents score as well from self-generated charts as from numeric output, yet plotting access alone does not translate into consistent use. Task slices show that charts lead on temporal, pattern, and anomaly reasoning, while text leads on precise reading and conditional verification. On three held-out benchmarks with four L L M backbones, method raises analytical accuracy and visual evidence quality over matched baselines.
Sources in this cluster
- Charts as Epistemic Actions for Experience-Driven Agentic Data Analysis
arXiv cs.AI · 22 Oct 2026, 03:26 AEST