APOD 0.1.0 – Agent Panel On Demand for Structured Multi Agent Dialogues - mauriceling/mauriceling.github.io GitHub Wiki

Citation: Ling, MHT. 2025. APOD 0.1.0 – Agent Panel On-Demand for Structured Multi-Agent Dialogues. Acta Scientific Computer Sciences 7(8): 10-13.

Link to [PDF].

Here is the permanent [PDF] link to my archive, link to the [repository], and link to the [video demonstration].

Panels of experts have long been used as mechanisms for reflection, debate, and consensus-building. However, convening such panels requires resources, coordination, and access to diverse expertise, which are not always feasible. With the rise of AI-based language models, it has become possible to simulate multi-agent panels that represent diverse perspectives in structured dialogues. Here, I introduce APOD (Agent Panel On-Demand) version 0.1.0, a Python-based framework that generates structured multi-agent dialogues among configurable AI personas. Each persona embodies a cognitive bias, communication style, and domain expertise. APOD supports reproducible, turn-based dialogues, logging in Markdown/Wiki formats, and moderator-driven synthesis. This paper describes the motivations, design, architecture, and a demonstration of APOD.