An exploration of n8n orchestration
and prompt engineering applied to
a project’s communications
FEATURED PAPER
By Luis Sanhueza B.
Santiago, Chile
- Summary
During the execution of large projects, high volumes of technically and temporally fragmented information are generated, dispersed in diverse formats (spreadsheets, PDFs, word processors, etc.) and physical locations. Although it is understood that the project manager and his work team can access all this information, paradoxically they cannot apply it to their decision-making as they would like.
Along these lines, the motivation arises to explore whether artificial intelligence tools and prompting methodologies can integrate diverse reports and generate prescriptive analyses of greater value. This exploration is materialized in a prototype of a PMO Agent that uses progress reports as a data source to deliver, on demand, integrated information and suggest preventive and corrective actions.
The Agent relies on platforms such as Hostinger[1], n8n[2], Supabase[3] and Postgres[4], and applies prompt engineering patterns (Persona, Disambiguation, System Simulator) to align its responses with PMBOK standards. Five critical dimensions were addressed: physical and cost projection, schedule evolution, project status, task tracking, and risk management—the latter incorporating RAG technology to maximize the veracity of content.
The exercise was developed by simulating a real estate construction program of 1,032 activities, complemented by 3 weekly reports of physical progress and risk monitoring.
The results obtained are encouraging in this exploratory context (simulated project, structured data, defined rules). The Prototypical Agent showed that it is technically feasible to integrate diverse and dispersed data (schedule, physical advances, risks) in a single source (PostgreSQL + vectors) and query them on demand in natural language; that using prompting patterns it is possible to detect inconsistencies in the information and alert on them (not just raw data), and that it is possible to cross management dimensions (such as schedule and risks) in a single query, generating a prescriptive analysis aligned with PMBOK.
But it is suggested to be cautious with the results. Far from replacing human communication, the PMO Agent protects it by reducing noise, ordering the flow of information, and freeing leaders to focus their energy on judgment, negotiation, and decision-making in context.
Statement on the use of Artificial Intelligence:
The author declares that artificial intelligence has been used exclusively as an object of technical study and for editorial support purposes for the grammatical optimization of the manuscript. The PMO Agent’s logical architecture, prompt engineering development, data analysis, and presented conclusions are the author’s original intellectual work.
- Fundamentals and Driving Idea of the exercise
The development of the “PMO AGENT” prototype is based on the need to address the communicational frictions that affect the success of projects. It is a practical laboratory to observe how AI can act as a bridge to transform raw and scattered data into strategic analysis.
More…
To read entire paper, click here
How to cite this paper: Sanhueza, L. B. (2026). AI Agent for PMO: An exploration of n8n orchestration and prompt engineering applied to a project’s communications; PM World Journal, Vol. XV, Issue V, May. Available online at https://pmworldjournal.com/wp-content/uploads/2026/05/pmwj164-May2026-Sanhueza-AI-Agent-for-PMO.pdf
About the Author
Luis Sanhueza B.
Santiago, Chile
Luis Sanhueza B. is an industrial civil engineer with a master’s degree in industrial engineering, with more than 35 years of experience in the management, direction and control of projects of varying complexity. Since 2007, he has been certified as a Project Management Professional (PMP®) by the Project Management Institute (PMI).
Throughout his professional career, he has specialized in project management and the implementation of Microsoft solutions for enterprise project management (EPM). Currently, he offers consulting services in industrial engineering and expertise, while researching the convergence of emerging technologies and project management. His main focus is on the development of prompt engineering architectures for cognitive support and strategic decision making in Project Management Offices (PMOs).
Luis can be contacted at lsanhba@gmail.com
[1] Hostinger provides optimized VPS infrastructure with 1-click installation template and official API node, allowing you to run n8n in a self-managed, private, and cost-effective way
[2] N8N is an open-source workflow automation platform that allows you to connect multiple applications and services to automatically execute tasks.
[3] Supabase is an open-source Backend-as-a-Service (BaaS) platform that allows developers to build applications quickly without having to manually configure and manage servers.
[4] Postgres is an open-source, object-oriented relational database management system (RDBMS) recognized for its high reliability, high performance, and ability to handle complex data loads. It uses SQL for queries and supports advanced features, ACID (atomicity, consistency, isolation, and durability) features, and is highly extensible.
