Search papers, labs, and topics across Lattice.
This paper introduces a pipeline for on-demand persona generation in multi-agent systems, enabling dynamic agent creation tailored to user characteristics and task demands. The key idea is to move beyond fixed agent architectures by crafting AI personas at runtime. The authors propose this approach as a way to improve efficiency and contextual appropriateness within agentic workflows.
Forget hard-coded agents: dynamically generated personas could unlock more efficient and personalized multi-agent workflows.
Recent advances in agentic AI are shifting automation from discrete tools to proactive multi-agent systems that coordinate multi-specialized capabilities behind unified interfaces. However, today's agent systems typically rely on hard-coded agent architectures with fixed roles, coordination patterns, and interaction flows that limit end-user personalization and make adaptation to individual needs and contexts difficult. Given this limitation, we argue that on-demand persona-based agent generation offers a promising path towards more efficient and contextually appropriate interaction within agentic workflows. By dynamically crafting agents and personas at run-time to match user characteristics, task demands, and workflow context, agentic platforms can move beyond one-size-fits-all configurations. We present a pipeline for on-demand persona generation in agentic platforms, detailing how real-time crafting of AI personas can be systematically integrated within agent systems, aiming to open new possibilities in agentic platform design paradigms.
