Constrained Process Maps for Multi-Agent Generative AI Workflows

Ananya Joshi, Michael Rudow · 2026

Read Constrained Process Maps for Multi-Agent Generative AI Workflows by Ananya Joshi, Michael Rudow (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-02034.txt.

Description

Ananya Joshi and colleagues offer bounded process maps for multi-agent workflows, framing the system as a Markov Decision Process with a one-way acyclic structure. Large language model agents perform complex workflows in strict settings, yet single-agent prompting hides doubt handling. Each agent fills a role or choice stage. Agent-level doubt uses Monte Carlo estimation, while system-level doubt ends there. A case study in AI safety review for self-harm spotting shows a 19 percent score increase and an 85 times human review cut.

Catalog record

CREATORAnanya Joshi, Michael Rudow
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE19232
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602.02034

Browse the full library · About the Tower of Babel