Case studyConstruction & BTPQ4 2025 – Q1 2026 · 4 monthsPublic procurement

DPGFast

Intelligent DPGF comparator — automated analysis and comparison of supplier price bids for public construction tenders

Discuss a similar project

Réponse courte

DPGFast is a tool developed by Digit-AI that automates the comparison of DPGF priced bills of quantities (Décomposition du Prix Global et Forfaitaire) for construction cost estimators. It imports the Excel files of templates and supplier responses, automatically detects differences (identical, modified, added, removed lines) through a multi-level matching algorithm and a Claude AI agent, then generates professional Excel and HTML exports with comparative statistics. The result: analysis time drops from 2–4 hours to under 10 seconds per lot (−95%).

Context

The challenges of bid analysis

In public construction tenders, comparing supplier price bids is a manual, lengthy and error-prone process

Construction cost estimators manually compare, line by line, Excel files of hundreds of items for each lot of each project. Every new supplier DPGF must be reconciled with the initial template to identify price gaps and added or removed lines — an entirely manual, time-consuming and high-error-risk process.

!

Manual Excel comparison

Cost estimators compare, line by line, Excel files of hundreds of items, with a high risk of human error.

!

Heterogeneous formats

Each project uses different formats (6 or 7 columns), making standardized processing impossible.

!

Added / removed lines

Suppliers modify the DPGF structure by adding or removing lines, making the comparison more complex.

!

Processing time

Several hours per lot for a manual analysis, multiplied by the number of suppliers and lots per project.

Functional scope

Construction cost estimators, project leads, purchasing management. Covered processes: DPGF file import, multi-supplier comparison, professional Excel and HTML exports, multi-lot batch processing.

Key constraints

Heterogeneous Excel formats (6 and 7 columns) · Merged cells and non-standard structures · Volume of hundreds of lines per file · Docker deployment compatible with Railway and on-premise.

Objectives

What DPGFast solves

Three areas of improvement to transform the bid-analysis process

Radical time savings

Cut analysis time from several hours to a few seconds per lot by automating the detection of differences between the template and supplier responses.

KPI: analysis time ÷100

Reliability and accuracy

Eliminate human errors through a multi-level matching algorithm (code, label, hierarchical context) and automated calculation validation.

KPI: 100% gap detectionKPI: 0 comparison errors

Professional exports

Automatically generate professional Excel and HTML exports, with comparative statistics, color coding and embedded comments.

KPI: standardized exportKPI: min/max/avg/spread stats
Features

What DPGFast enables

A set of features covering the full DPGF comparison cycle, from import to export

📊

Multi-format import

Support for 6-column and 7-column DPGF formats with automatic format detection. Adaptive AI agent for non-standard files.

🔍

Intelligent comparison

Multi-level matching algorithm: section code, line number, label (exact, partial, semantic) with hierarchical context constraints.

📑

Professional Excel export

Excel files with professional design: embedded logo, merged headers, simplified supplier columns and statistics (min, max, average, spread).

🌐

Interactive HTML view

Web comparison interface with tooltips on modified fields, color legend and smooth navigation between lots and suppliers.

🤖

Claude AI agent

Adaptive Excel-reading agent using Claude Tool Use to analyze complex Excel structures (merged cells, non-standard formats).

📁

Multi-lot batch processing

Automatic import and processing of complete projects with multiple lots, buildings and suppliers through an automated CLI pipeline.

Stack

Tech Stack

A modern Python stack optimized for Excel file manipulation and cloud deployment

Backend

Python 3.8+FastAPIUvicorn

Frontend

HTML5 / CSS3TailwindCSSHeroicons

Excel & Data

openpyxlpandasxlrd

Artificial intelligence

Claude (Anthropic)Tool Use API

Database

SQLite (dev)PostgreSQL (prod)

Infrastructure

DockerRailwayGitHub

Differentiating pattern — multi-level matching: DPGFast's comparison algorithm is not limited to matching by line number. It combines section code, line number, exact label, partial label, semantic similarity and hierarchical context to precisely identify identical, modified, added or removed lines — even when the supplier has restructured the document.

Architecture

Architecture Diagrams

Modular backend-oriented architecture with CLI and Web interfaces, a central comparison engine and adaptive AI agents

APPLICATION ARCHITECTURE — DPGFAST v2.5.0USERCost estimatorConstruction cost estimatorCLI INTERFACEprocess_folder.pyBatch import, multi-lotWEB INTERFACEFastAPI + HTML/CSSUpload, view, exportAI AGENTSClaude Tool UseAdaptive Excel readingCORE ENGINEcompare.pyComparison enginedatabase.pyData access layerfile_validator.pyFile validationEXPORT ENGINEexcel.pyhtml_view.pyhtml_db.pyhtml_table.pyDATABASESQLiteDevelopmentPostgreSQLProductionVALIDATORSmodel.pysupplier.pycalculations.pycomparison.py.xlsx.htmlExcel files
Results

Measured impact and gains

The gains observed after deploying DPGFast at our clients

−95%
Analysis time per lot
2–4 h → < 10 s
100%
Gap detection
No line missed
2
Supported formats
6 and 7 columns + auto-detection
v2.5
Version in production
Deployed on Railway
IndicatorBefore DPGFastWith DPGFastGain
Time per lot (5 suppliers)2–4 hours (manual)< 10 seconds−95%
Human error riskHigh (fatigue, inattention)None (automated)Eliminated
Detection of added/removed linesManual, often incompleteAutomatic and exhaustive100%
Export qualityManual formattingAutomatic professional designStandardized
Multi-lot processingLot by lot, manualAutomatic batchAutomated

Qualitative gains

💡

Informed decisions

Comparative statistics (min, max, average, spread) enable an objective and fast analysis of supplier bids.

🧠

Focus on analysis

Cost estimators focus on interpreting the gaps rather than on manual, line-by-line comparison.

Full traceability

Versioning of comparisons, change history and comments embedded in the exports.

🔒

Data security

File validation on upload, protection against SQL injection, secure session-based authentication.

📈

Scalability

Modular architecture allowing new DPGF formats, export types or AI agents to be added without an overhaul.

🎨

Professional image

Professionally designed exports with logo, color palette and consistent layout for each client.

Let's start by talking it through

No aggressive sales pitch. No 12-step form. Just 30 minutes to understand your situation and see if we can help. First conversation free, no strings attached.