Agentopic, an agent-based workflow using LLMs, achieves F1-score of 0.95 on BBC dataset topic modeling with full explainability of reasoning.

Agentopic uses multiple AI agents that collaborate to identify topics, validate them, organize them hierarchically, and explain them in natural language, enabling users to trace how topics are assigned.

When seeded with topics from the BBC dataset, Agentopic achieves an F1-score of 0.95, matching GPT-4.1, improving on LDA (0.93), and close to BERTopic (0.98). Unseeded, it generated 2045 semantically coherent topics organized across six hierarchical levels, expanding the original five-category structure.

By embedding explainability throughout the workflow, Agentopic offers an interpretable alternative to black-box models, particularly valuable for applications in finance and healthcare where reasoning transparency is critical.