This repository serves as the research companion for Ask SURI, an agentic AI architecture that integrates Large Language Models (LLMs) with automated tool-calling capabilities to automate construction schedule intelligence.
The construction industry relies heavily on complex scheduling methodologies, primarily the Critical Path Method (CPM) and Earned Value Management (EVM), managed within specialized software like Oracle Primavera P6. However, extracting actionable insights from these schedules remains a highly manual, time-consuming process.
Ask SURI bridges the gap between natural language queries and structured schedule data. It employs a decoupled client-server architecture that parses native Primavera P6 XER files, executes CPM passes, computes DCMA 14-point schedule assessment metrics, and calculates Earned Value indices. An LLM orchestration layer equipped with domain-specific tools enables the model to retrieve targeted schedule data and stream results via Server-Sent Events (SSE).
The system is designed around three core subsystems:
- Data Processing Pipeline: XER file ingestion and schedule metric computation.
- AI Orchestration Layer: LLM tool calling and narrative generation.
- Real-Time Communication Layer: Server-Sent Events (SSE) streaming.
(See the /architecture folder for detailed diagrams and component descriptions.)
- Native XER Parsing: Direct ingestion of Primavera P6 schedule files.
- Automated CPM Calculation: Forward and backward passes to determine early/late dates and total float.
- DCMA 14-Point Assessment: Automated schedule health scoring based on industry standards.
- EVM Computation: Calculation of SPI, CPI, SV, and CV.
- Domain-Specific Tool Calling: 8 specialized tools enabling the LLM to query the schedule database.
- Conversational Interface: Natural language interaction with complex project data.
This repository contains architectural documentation and code samples demonstrating the core methodologies used in the Ask SURI system. It does not contain the full production codebase.
/architecture: System diagrams and design documentation./code_samples: Snippets demonstrating XER parsing, CPM calculation, and LLM tool definitions./paper: The academic manuscript detailing the research and evaluation.
If you use this architecture or methodology in your research, please cite the accompanying paper:
@article{waidande2026asksuri,
title={Agentic Schedule Intelligence: Integrating Large Language Model Tool Calling with Critical Path Method Analysis for Automated Construction Project Controls},
author={Waidande, Sushant},
journal={Journal of Computing in Civil Engineering},
year={2026},
note={Under Review}
}The documentation and code samples in this repository are licensed under the MIT License. See the LICENSE file for details. The Ask SURI production application is proprietary software.