- Bricks & Bytes Bulletin
- Posts
- Lupa Technology: AI Document Analysis With An Unconventional Approach
Lupa Technology: AI Document Analysis With An Unconventional Approach
Want to get your message in front of 2,011 highly engaged innovation leaders? Check out our sponsorship offers.
INDUSTRY INSIGHTS
Lupa Technology: AI Document Analysis With An Unconventional Approach
TL;DR
Lupa Technology is a construction dispute resolution platform founded by a Turner Construction veteran. The company uses AI to analyze project documents including emails, drawings, schedules, and handwritten notes. The system performs document search, semantic tagging, multilingual analysis, and drawing comparisons to identify discrepancies in construction projects.
The platform processes documentation for both active projects and dispute claims, reducing the time required for document review from weeks to hours.
Note: This post is in partnership with Lupa Technology.
Having worked as a Quantity Surveyor (no, US folks we don’t just count bricks!) for a number of years and doing my fair share dispute resolution - and the laborious process of trawling thousands and thousands of documents to find small nuggets of evidence - I found this company - Lupa technologies - tackling what might just be one of the biggest pain points for anyone working in the claims side of the construction industry.
The problem - and this isn’t just related to claims - is that the construction industry generates unprecedented volumes of data yet most organizations struggle to extract meaningful insights from the millions of documents, communications, and technical files that accumulate across project lifecycles.
This problem is not novel and has resulted in a new category of AI-powered platforms emerging to address the challenge of extracting meaningful data from documentation. However, what is interesting is Lupa Technology’s focus on dispute resolution as a starting point.
Founded by Vladimir Milovanovic, a former Turner Construction "fixer" who spent two decades troubleshooting problematic projects globally, Lupa addresses a fundamental problem: the manual impossibility of analyzing the 2-3 million documents that accumulate on average construction projects. The platform transforms this data deluge into structured intelligence for both live project controls and post-construction dispute resolution.
What Lupa Technology Does
Lupa Technology is an AI-powered SaaS platform that consolidates fragmented construction project data into a single, searchable environment. At its core, the platform solves a practical problem: construction teams typically store project information across dozens of different systems, such as emails in Outlook, drawings in document control systems, schedules in specialized software, communications in messaging platforms, and reports scattered across various folders and drives.

The initial extract report provides teams with their first comprehensive view of project data, identifying what information is available and what might be missing.
The platform ingests this disparate data, processes it using construction-trained AI models, and provides tools for searching, analyzing, and reporting across the entire dataset.
Rather than requiring teams to open multiple applications and manually piece together information, Lupa creates a unified intelligence layer that can answer questions, identify patterns, and generate insights from the complete project record.
Lupa serves two primary markets: live project controls during construction and forensic analysis for dispute resolution.
During active construction, teams use the platform to monitor project health, identify emerging issues, and make data-driven decisions. For claims and disputes, legal teams and consultants use Lupa to rapidly analyze historical project data and build evidence-based cases.
Core Platform Features
Data Integration and Processing
Connects to over 50 different data platforms including Microsoft ecosystems, project management systems (Procore, Aconex), and communication platforms
Processes data at approximately 5 minutes per gigabyte
Supports multilingual analysis across 115 languages
Handles all document types including PDFs, emails, images, schedules, and handwritten notes through OCR
AI-Powered Search and Analysis
Semantic search that understands context and intent beyond keyword matching
Bulk tagging that automatically categorizes content according to contractual provisions from FIDIC, AIA, and NEC contracts
Sentiment analysis trained specifically on construction communications
Deep research reports that analyze 100% of document collections
Real-time chat interface for instant queries against project data
AI power queries and prompts for complex Excel tables with the capability to obtain references from batches of unstructured documents
Specialized Construction Analytics
Drawing comparison tools that automatically overlay and analyze changes between revision sets
Schedule analytics supporting Primavera P6, Asta PowerProject, and Microsoft Project
Communication analysis across emails, meeting minutes, reports, and messaging platforms
Photograph analytics using computer vision to identify objects and extract metadata, plotting photos with geolocation coordinates on a map
Automated weekly reporting that tracks issues by category, location, and responsible parties
AI powered Contract Reviews that generate CRS (Contract Review Sheets), Contract Risk Assessment Reports, etc.
Reporting and Intelligence
Pre-built workflows for common construction processes (delay analysis, change order tracking, risk assessment, RFI prioritization, Rebuilding As-built timelines, etc.
Customizable dashboards showing project metrics and data composition
Export capabilities to Excel, Word, PDF and other standard formats
Timeline reconstruction for forensic analysis
Evidence compilation tools for claims and dispute resolution
Platform Architecture and Core Functionality
Lupa operates as a cloud-based SaaS platform that ingests data from over 50 different construction and business systems. The architecture centers on two primary processing approaches, each optimized for different analytical requirements.

The main dashboard displays data processing metrics and composition analysis across integrated project sources. Dashboards are tailored for two main use cases:
1. Provide to users a clear understanding what data has been ingested by Lupa and what data may be missing – to inform the user if additional data should be provided for completeness
2. Allows users to drill down into project issues and understand them, by incrementally increasing the level of detail, providing bite-size chunks of critical information on topic
AI architecture allows different approaches: The first approach, Retrieval Augmented Generation (RAG), powers real-time chat interfaces by chunking documents and retrieving relevant sections for immediate AI interpretation. This method delivers responses within seconds but sacrifices comprehensiveness for speed; suitable for quick queries during active project management.
The second approach, termed "Deep Research Reports," processes entire document collections line-by-line to generate comprehensive analyses. While requiring 15-20 minutes for completion, these reports analyze 100% of available data rather than selected chunks. According to Vlad, "If you have 10,000 emails to process and produce a report that tells you something about a particular problem, what I expect is not to receive that report in seconds. I'm perfectly okay to receive it in 15, 20 minutes, as long as it's fully comprehensive."
Data processing occurs at approximately five minutes per gigabyte, enabling analysis of multi-terabyte datasets within days rather than the months required for manual review. The platform maintains enterprise-grade security through encryption, access controls, and certified development practices - critical considerations given the sensitive nature of construction project data.
Speed Vs Thoroughness
The dual processing approach solves a fundamental trade-off in construction data analysis. Project managers need immediate answers during active construction: "Who approved the change to the HVAC design last week?" - which RAG delivers instantly. But when investigating a delay claim worth millions, teams need comprehensive analysis that doesn't miss critical evidence buried in thousands of documents.
Consider a dispute over project delays on a $200 million hospital construction project. Traditional manual review might take a team of analysts 6-8 weeks to examine relevant communications and schedules - and they will work off sample documentation, not the entire data set Lupa's deep research processing can analyze the entire dataset in 3-4 days, then enable rapid follow-up queries as the case develops. The speed advantage becomes multiplicative; faster initial analysis enables more thorough investigation within the same timeframe, with greater accuracy
Document and Communication Analysis
The platform's text analysis capabilities extend beyond simple keyword searching to semantic understanding trained specifically on construction terminology. Users can query concepts like "material defects" and receive contextually relevant results even when documents use different terminology or a different language.

The semantic search function demonstrates contextual understanding, surfacing relevant results across document types and languages.
Bulk tagging automatically categorizes content according to contractual provisions from FIDIC, AIA, and NEC contracts known to cause construction issues. This automated classification leverages decades of industry knowledge to identify potential risk areas without manual document review.
Communication analysis processes emails, meeting minutes, reports, and even messaging platform content through OCR technology for handwritten notes. Sentiment analysis, trained on construction contexts, identifies communications indicating emerging problems or disputes, often providing early warning signals before issues formally escalate.
The platform supports multilingual analysis across 115 languages, essential for international projects where communications occur in multiple languages. This capability enables semantic searching across language barriers, finding conceptually related content regardless of the source language.
Finding Needles In Haystacks
Traditional keyword searches fail when critical information is described using different terminology. Searching for "concrete defects" might miss emails discussing "poor workmanship in structural elements" or "quality issues with pour #47." Lupa's semantic understanding connects these conceptually related discussions.
In practice, this capability recently helped a claims team identify a pattern of quality issues across a project. While individual incidents were described differently e.g. "surface irregularities," "finish problems," "cosmetic concerns", the semantic analysis revealed they all related to the same subcontractor's work.
The multilingual capability becomes essential on international projects. A dispute on a Middle East project involved Arabic site reports, English engineering communications, and Hindi labor correspondence. Lupa's ability to search concepts across all three languages enabled comprehensive case development that would have been practically impossible with manual translation and review.
Technical Drawing and Schedule Analytics
One of Lupa's more specialized capabilities addresses design change management through automated drawing comparison. The platform creates drawing registers from PDF files, extracts vector and text information, then performs bulk overlay comparisons between revision sets.

The drawing analytics interface shows overlay comparisons with AI-generated change narratives identifying specific modifications.
The AI generates detailed narratives describing changes between drawing versions. Vlad cited an example where the platform detected a single-letter change in fire rating codes (from "E 90" to "EI 90") that represented a three-million-euro cost impact. Industry research suggests 30% of design changes go unnoticed in traditional review processes, making automated detection particularly valuable for risk management.
For schedule analysis, the platform imports data from Primavera P6, Asta PowerProject, and Microsoft Project to perform over 30 different analytical functions. Rather than simply displaying schedule data, the system feeds these analytics to AI models that generate reports explaining schedule performance, critical path changes, and potential delay impacts.
All the construction data is in one place, and Lupa bridges data fragmentation across multiple storage providers (onedrive, sharepoint, dropbox, email…) as well as providing integration across data silos (text, drawings, schedules, photos…). Drawing changes and Schedule changes typically tell you what happened, whereas text provides context and understanding why it happened.

Weekly automated reports break down project issues by category, location, and responsible parties based on document analysis.
Catching Critical Changes
Manual drawing comparison is notoriously error-prone and time-consuming. A design manager on a typical commercial project might receive 50-100 revised drawings weekly. Traditional "red-line" comparison requires printing both versions and manually checking for differences… a process that typically catches only obvious changes while missing subtle but critical modifications.
The fire rating example illustrates the business impact. The single-letter change from "E 90" to "EI 90" altered door requirements from basic fire resistance to integrity protection, requiring different hardware, installation methods, and potentially affecting the building's fire safety certification. Missing this change could have resulted in non-compliance issues, rework costs, and schedule delays far exceeding the three-million-euro direct cost impact.
For schedule analytics, consider a project experiencing delays across multiple work packages. Traditional schedule analysis requires manually comparing dozens of schedule updates to identify patterns and root causes. Lupa's automated analysis can instantly identify that delays consistently follow design approval bottlenecks, enabling project teams to address the systemic issue rather than just individual delayed activities.
Real-World Application Scenarios
Lupa's capabilities translate into several distinct use cases across the construction project lifecycle.
During active projects, teams use the platform for real-time project controls, monitoring communications for sentiment indicators that suggest emerging problems, and tracking design changes that might impact schedule or budget.
The weekly reporting functionality automatically dissects project data into chronological segments, identifying and categorizing issues as they surface in project communications. This enables proactive problem-solving rather than reactive crisis management.
For post-construction disputes, the platform serves as a forensic analysis tool. Claims consultants and legal teams use Lupa to establish project chronologies, identify relevant evidence from vast document collections, and build fact-based narratives for arbitration or litigation. The platform's ability to cross-reference information across different data types eg linking an RFI to its corresponding drawing and schedule impact, provides comprehensive case development support.
One documented case involved processing over 900GB of email data (more than one million messages) in two days to identify critical evidence that had eluded traditional discovery methods for months. The comprehensive analysis approach often reveals connections and patterns that manual review cannot practically achieve.
Prevention vs. Crisis Management
The value difference between proactive and reactive project management compounds dramatically over time. Weekly automated reports can identify emerging issues when they're still manageable. For example, sentiment analysis might detect increasing frustration in subcontractor communications about material deliveries weeks before formal delay notices are submitted.

Schedules report
Industry Integration and Workflow Considerations
Lupa integrates with established construction technology ecosystems through APIs and direct data uploads. Common integrations include*:
Microsoft environments (OneDrive, SharePoint, Teams, Outlook),
Project management platforms (ACC, Onsite, ThinkProject, Procore, Aconex), and specialized construction software.
Cloud storage (Dropbox, Azure)
Messaging platforms (WhatsApp, Viber)
*A selection of some of the 50+ integrations
Typically, a key requirement for any large organization considering purchasing a software is the ability for it to integrate into their existing system. The big players (ACC, Procore, Aconex… etc) always come up.
The platform includes pre-built workflows that mirror common industry processes such as delay analysis, change order tracking, and risk assessment. These workflows embed expert knowledge directly into the software, guiding users through complex analyses that traditionally require specialized consulting support.
Training requirements vary by user role and application complexity. Basic AI-driven reports and search functions can be utilized immediately, while advanced features like custom report generation and complex data analysis may require several weeks for full proficiency. The platform provides templates and guided workflows workflows and Help documentation with a Chatbot to accelerate adoption across different skill levels.
Technical Limitations and Considerations
While comprehensive, Lupa faces certain technical constraints common to AI-powered platforms. The accuracy of semantic search and automated categorization depends on training data quality and ongoing model refinement, on one hand and User’s ability to provide right constraints on the other. The nature of forensic analysis demands a human-in-the-loop and while the platform could move far more quickly on its own, workflows are split into stages where users can validate results at each step. In practice, Lupa’s speed and accuracy are limited by the collaborative effort of the AI and User working together. Processing time scales with data volume, making real-time analysis impractical for very large datasets during peak processing periods. Organizations must balance comprehensive analysis with time constraints, particularly for urgent project decisions.
The platform's effectiveness depends heavily on data quality and completeness. Fragmented or poorly organized source data can limit analytical accuracy, requiring organizations to maintain consistent data management practices across integrated systems - Lupa will not be aware of documents that are not in the system.
Integration complexity varies significantly depending on existing technology infrastructure. Organizations with established data governance and standardized systems typically achieve smoother implementations than those with fragmented or legacy technology environments, while on the other hand, those with fragmented data environments will gain most value from the system.
Market Position and Competitive Dynamics
Lupa occupies a unique position between traditional construction project management software and legal e-discovery platforms. Unlike generic business intelligence tools, the platform incorporates construction-specific AI training and workflows that understand industry terminology and processes.
Compared to traditional e-discovery software used in legal contexts, Lupa handles engineering data types (drawings, schedules, technical specifications) that general litigation support tools cannot process effectively. This specialization provides advantages for construction disputes that involve both legal and technical analysis.
The platform differs from conventional construction management systems by focusing on analytics and intelligence rather than daily workflow management. Organizations typically maintain existing project management platforms while using Lupa as an analytical overlay for deeper insights.
Pricing is based on data volume rather than revenue or user count, with monthly, annual, and enterprise licensing options. The cost-plus approach for infrastructure and AI processing provides predictable expenses that align with project-based business models common in construction.

A selection of some of Lupa’s clients
Implementation and Organizational Impact
Successful Lupa implementation requires consideration of both technical and organizational factors. Data preparation and integration setup typically require 2-4 weeks depending on source system complexity and data volume. Organizations with standardized data practices and modern technology infrastructure generally experience smoother implementations, others may have to process more data to ensure everything is on the platform.
User adoption varies by role and technical comfort level. Project managers and claims specialists often adapt quickly to search and reporting functions, while advanced analytics capabilities may require additional training and change management support.
The platform's impact on organizational workflow depends largely on implementation scope and user engagement. Organizations using Lupa for specific applications (such as claims analysis) report immediate value, while those attempting comprehensive project intelligence transformation require longer adoption periods and more extensive change management.
Finally, Lupa holds some of the highest security credentials includes ISO 27000, SOC2 and “Highest Pen-Test Rating” by Illumant.
Conclusion
If you believe that your data holds the information necessary to fight claims or you want to do more data-driven decision making, Lupa has found an unique niche in construction industry for you. For anyone who has spent sleepless nights manually combing through plethoras of project documents or scrolling endlessly through email threads searching for that one critical piece of evidence, Lupa Technology might just hold the answer.
The platform's emergence reflects a broader maturation in construction technology, moving beyond simple digitization toward genuine intelligence. Where previous generations of construction tech focused on replacing paper with screens, Lupa tackles the more complex challenge of making sense of the digital chaos that digitization created.
Perhaps most significantly, Lupa democratizes access to sophisticated analytical capabilities that were previously available only to the largest organizations with dedicated forensic teams. A mid-size contractor can now perform the same level of comprehensive dispute analysis that once required armies of consultants and months of billable hours.
For construction professionals evaluating whether platforms like Lupa represent worthwhile investments, the value proposition is straightforward: transform the economics of evidence discovery. Instead of teams spending weeks manually reviewing documents to build a timeline of events, that same analysis can be completed in days, freeing up experienced professionals to focus on strategy, negotiation, and resolution rather than document archaeology.
With solutions like Lupa, the construction industry, notorious for its litigious nature and black-hat claims culture, may finally be able to focus on doing what it does best. Build!
WEEKLY MUSINGS
Snaptrude AI, Safety, Nuclear Power
Mark your calendars
The most important commitment
Back in the global spotlight
SPECIAL OFFERS
The GTM Bundle
Together, they give you the full GTM playbook for construction tech from building trust-first marketing to structuring your sales motion, pricing models, and land-and-expand strategy.
![]() | ![]() |
Individually, each guide is $100. Today, you can get both for $150.
YOU MIGHT ALSO LIKE
Premium Insights
More Insights
Reports and Case Studies
Most Popular Episodes
Super Series
OUR SPONSORS
![]() | Archdesk - The #1 construction management software for growing companies. Manage your projects from Tender to Handover. |
![]() | BuildVision - Streamlining the construction supply chain with a unified platform for contractors, manufacturers, and stakeholders. |