Agents Thinking Fast and Slow: A Talker-Reasoner Architecture

TL;DR

This paper introduces a modular 'Talker-Reasoner' architecture for AI agents, inspired by human fast and slow thinking, to improve conversational and reasoning capabilities with reduced latency.

Contribution

It proposes a novel two-system architecture separating fast conversational response generation from slow reasoning and planning, enhancing modularity and efficiency.

Findings

Demonstrated in a sleep coaching agent application

Reduced response latency compared to monolithic models

Improved modularity and clarity in agent design

Abstract

Large language models have enabled agents of all kinds to interact with users through natural conversation. Consequently, agents now have two jobs: conversing and planning/reasoning. Their conversational responses must be informed by all available information, and their actions must help to achieve goals. This dichotomy between conversing with the user and doing multi-step reasoning and planning can be seen as analogous to the human systems of "thinking fast and slow" as introduced by Kahneman. Our approach is comprised of a "Talker" agent (System 1) that is fast and intuitive, and tasked with synthesizing the conversational response; and a "Reasoner" agent (System 2) that is slower, more deliberative, and more logical, and is tasked with multi-step reasoning and planning, calling tools, performing actions in the world, and thereby producing the new agent state. We describe the new Talker-Reasoner architecture and discuss its advantages, including modularity and decreased latency. We ground the discussion in the context of a sleep coaching agent, in order to demonstrate real-world relevance.