-
Digital as-built via 3D scanner as the basis for transparent AI quality assurance
-
AI-supported scan analysis for automated labeling and lightning-fast billing
-
Human-in-the-loop: Human control guarantees complete legal certainty
-
Real-time collaboration via web platform with live comments directly on the construction site
-
Efficiency boost & risk minimization: fewer rectifications, faster project completion
Digitalization is shaping almost every area of the construction industry today - especially in FTTH projects. What used to be done with a red pen and on paper is now a matter of modern technology: automated as-built documentation is becoming the digital foundation for a transparent, traceable and billing-relevant presentation of construction work. It's not just about making progress on the construction site visible, but also documenting it in such a way that it can be processed and checked quickly - for faster project completion, fewer rectifications and more clarity.
Process optimization with AI
According to Forbes Advisor, over 50% of companies are already using AI specifically for process optimization - and the trend is rising. The proportion is also growing in Germany: according to ZEW Mannheim, around 12% of companies actively use AI.
The technology has therefore long been more than just an experiment - it has become part of functioning processes. However, quality assurance on the construction site in particular shows that
AI alone is not enough.
The requirements for FTTH construction documentation are particularly high. Documentation not only forms the basis for subsequent billing, but also serves as proof that work has been carried out correctly and ensures network quality. If data is missing or inaccurate, there is a risk of rework, unnecessary costs and discussions about responsibilities. This is exactly where DeepUp comes in: Our 3D scanner captures high-resolution image data during the construction process, which is processed in the shortest possible time and enriched with quality-assured information. In addition to billing-relevant data, this also automatically generates survey data - without any additional effort.
We use AI-supported data processing to turn raw data into usable information. The artificial intelligence analyzes scans, cleans and sorts image material, labels relevant elements and prepares the results for billing. These processes run automatically and quickly, are scalable and ensure high efficiency - especially for extensive projects with many house connections.
AI will never work without human control
However, even if the advantages are obvious, AI remains a tool - and no substitute for human control. This is because automated systems reach their limits in practice. Distinguishing between a ball marker and a construction site helmet can be just as challenging for an AI as poorly lit images, unfavorable camera angles or minimal visual differences - such as cables with seven instead of nine wires. A recognition rate of 80% sounds good, but is not enough when it comes to legally compliant documentation and correct invoicing of construction work.
That's why we at DeepUp deliberately pursue a human-in-the-loop (HITL) approach. Our AI brings speed and structure to raw data processing, but humans remain the decisive quality anchor. Trained specialists check borderline cases, correct results if they are ambiguous and provide the necessary contextual understanding - an aspect that no algorithm can replace. At the same time, this manual data review also improves the AI itself: Every piece of feedback flows into further development, every correction helps the system to learn.
How does HITL work at DeepUp?
Practical experience on the construction site shows: Time is often a critical factor. Scans have to be carried out directly in the trench and subsequent improvements are often not possible. At the same time, construction site personnel are not necessarily specialized in documentation. We offer a simple and visual solution so that everyone involved can still work together: with our construction monitoring platform, we provide scans in an intuitive web view - including a comment function for site managers or project managers. This allows queries to be clarified directly in the tool, context to be added and any discrepancies to be quickly resolved.
In the end, it becomes clear that quality assurance with AI is only really strong when humans and machines work together. The combination of automated efficiency and human accuracy ensures that not only theoretical planning data is fulfilled - but that real, reliable information is available. DeepUp stands for this new form of network documentation: digital, fast, verifiable - and always with a clear commitment to quality.
