Abstract
In the context of multi-agent artificial intelligence (AI) environments in which a supervisor large language model (LLM) agent participates in a bounded collaborative reasoning session alongside a pre-declared team of specialist LLM agents, a system is proposed herein that dynamically spawns agents on demand by allowing the supervisor LLM agent to emit structured spawn directives during selected phases of a session. The team of specialist LLM agents can thereby expand in response to domain gaps discovered during the session while remaining subject to bounded-admission and guaranteed-termination constraints.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Sáez, Antonio Gallego and Cazila, Juan, "REACTIVE MID-SESSION EXPANSION OF A MULTI-AGENT LARGE-LANGUAGE-MODEL TEAM UNDER BOUNDED ADMISSION AND GUARANTEED-TERMINATION PHASE CONSTRAINTS, WITH CROSS-SESSION AGENT REINCARNATION", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/10946