Skip to main content
PHÚ QUÂNConstruction · Workflows · AI

Phu Quan · Construction in Vietnam

Developing AI workflows and agents for construction and business operations

Phu Quan is a Vietnam-based construction company developing AI workflows and agents for construction documentation, project coordination, and business operations. Our development is grounded in practical industry needs, with human review at key decision points.

Explore our development focusDiscuss a collaboration

Practical needs we aim to address

  • Documents are spread across Word files, Excel spreadsheets, PDFs, and drawings.
  • The same information is entered repeatedly into different forms.
  • Consolidating quantities, updating progress, and preparing reports take significant time.
  • Document versions and information from different sources are difficult to track and compare.
  • Engineering, site, and management teams need coordinated information.

Four areas of research and development

Each development direction starts with a specific task, defined inputs, useful outputs, and a responsible reviewer.

01Development direction

Document processing workflows

Our goal is to receive documents, extract information, cross-check data, prepare drafts using templates, and route results to the responsible reviewer.

Intended users
Technical staff and document controllers.
Input data
Word files, Excel spreadsheets, PDFs, templates, and authorized reference documents.
Supported tasks
Extraction, cross-checking, identifying missing information, and drafting.
Outputs
Information tables, review checklists, and template-based drafts.
02Development direction

AI agents for work support

We are exploring assistants that receive requests, search authorized sources, use tools, and carry out related steps to prepare results that people can verify.

Intended users
Engineering, coordination, and business operations teams.
Input data
Work requests, authorized documents, and defined tool permissions.
Supported tasks
Information retrieval, planning processing steps, drafting, and consolidating results.
Outputs
Drafts or reports with supporting references and decisions flagged for human review.
03Development direction

Project management workflows

Our goal is to consolidate progress, quantity, and quality information, track tasks, and assist with periodic reporting.

Intended users
Project managers and site teams.
Input data
Schedules, quantity tables, site diaries, inspection records, and site reports.
Supported tasks
Consolidating updates, tracking tasks, and identifying information requiring cross-checks.
Outputs
Draft progress reports, task registers, and issue lists.
04Development direction

Data and tool integration

We are researching connections between documents, spreadsheets, and existing management tools, prioritizing workflows that are practical to use and can grow over time.

Intended users
Operations and data management teams.
Input data
Documents, spreadsheets, data structures, and granted access permissions.
Supported tasks
Standardizing data, reducing repeated entry, and exploring information exchange between tools.
Outputs
Shared data structures and integration proposals for validation before deployment.

From source documents to reviewed results

An example of a workflow to support construction documentation.

  1. 01Receive Word / Excel / PDF files
  2. 02Extract information
  3. 03Cross-check and flag items for review
  4. 04Prepare a template-based draft
  5. 05Responsible reviewer approves
  6. 06Export the result

This is an illustrative planned workflow using sample data, not a live product. Extraction, cross-checking, and drafting steps represent our development direction.

Planned use of Claude

Phu Quan plans to investigate the use of Claude for document understanding, information extraction, drafting support, and coordinating processing steps through tools. Any integration will be assessed against the needs of each workflow, output quality, and operational effectiveness.

Using AI tools to support work is different from integrating AI into a product. We have not announced a Claude API integration, deployed products, or customer adoption.

  • Provide supporting references where appropriate and flag uncertain information.
  • Keep human review at key decision points before results are used.
  • Limit data and tool access to the scope explicitly authorized.

Discuss a collaboration

Discuss your practical needs, document types, and work scope so we can identify a suitable direction together.

Contact