Document Type
Poster
Publication Date
Fall 2025
Abstract
Project management apps often center around tracking tasks but in reality, setbacks come from system-level dynamics (handoffs and review gates, constrained specialists with overloaded schedules, etc), which quietly compound until nothing ships on time. This work presents PULSE, an agentic workflow optimization system that turns event logs and operational data into a three-component manager-ready report including 1. skill- and capacity-aware task assignments, 2. bottleneck detection from event logs, and 3. human-in-the-loop AI-assist recommendations. I evaluate my MVP using a synthetic dataset built with Ollama, mimicking cross-functional collaboration in a mid-sized e-commerce company. The data illustrates realistic departments, projects, tasks, and event logs into which the algorithm intentionally encodes six delay patterns and handoff structures for benchmarking. Future work will add calendar integration, dynamic project/task updating, continuous re-evaluation of workflow efficiency, and a web interface + Slack/Teams chatbot for intuitive day-to-day use by real teams.
Recommended Citation
Ng, Tiffanie, "PULSE: An Agentic Workflow Optimization Tool for Cross-Functional Teams Task Assignment, Bottleneck Detection, and Efficiency Recommendations" (2025). IPHS 391: Interdisciplinary AI Frontiers. Paper 11.
https://digital.kenyon.edu/dh_iphs_391/11
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.