AI-supported construction calculation assistant

Calculate faster and win the right tenders

The challenge

For the bid submission in the construction industry, long lists of requirements are painstakingly compiled in text form with calculations for the required craftsmen’s, machinery and material services. This construction calculation is done by experienced professionals whose time is very valuable for the company.

BRZ Deutschland GmbH

BRZ Deutschland GmbH is a specialist in organisation and construction IT. With organisational consulting, an ERP construction software platform, outsourcing services and training offers, BRZ supports companies in the construction industry in both digitising processes and organising them in a cost-saving and efficient manner.

How we helped

Since many elementary requirements for a construction object are often repeated in the various tenders, the XITASO data science experts developed a solution to make this calculation process more efficient. Based on past tender calculations and with the help of Data Science & Artificial Intelligence methods, BRZ 365 AI calculation saves calculators a lot of time with precise recommendations. This can now be used for complex requirements of the construction project.

Technologies

C#/.Net Blazor Docker Terraform Vektordatenbank Python Natural Language Processing (NLP) Azure Machine Learning

Winning tenders with BRZ 365 KI calculation

The Construction Calculation Assistant uses intelligent algorithms to calculate suitable proposals for bid items. The addition to the BRZ 365 calculation platform thus saves calculators a lot of time in time-consuming routine work.

Experience values and calculation approaches for new tenders can be taken over from the data pool of already calculated offers – even in the case of differently worded texts with the same function descriptions.

Calculation proposals on three levels

With BRZ 365 KI-Calculation, calculators always have full control over which past projects are included in the dataset and which automation level is used when processing the so-called bill of quantities:

1. Recommendations of individual focus positions

Construction calculators process item by item and have a series of proposals generated for each item. They always decide for themselves which proposals are accepted and which are not.

2. Classification of the bill of quantities

The classification of individual items visualizes the respective contribution in relation to the total price. This allows cost calculators to focus on expensive items and let the AI calculate the low-cost items.

3. Fully automatic calculation of all positions

The bill of quantities, including surcharges, prices, article and equipment lists, is calculated partially or completely automatically. As a result, construction calculators benefit from maximum time savings.

“Thanks to XITASO’s data science expertise, we were able to bring an innovation to the market with the AI calculation that gives BRZ a great competitive advantage and users* real added value.”

Johannes Gunkel
Product Owner
BRZ Deutschland GmbH

The XITASO Data Quick Check

Validation of the project vision in only 15 days

BRZ’s vision was based on the assumption that the required craftsman, equipment and material services are similar if the requirements are very similar in content. To test this hypothesis, requirement texts were compared using Natural Language Processing (NLP) and the respective prices were controlled. This ensured feasibility before resources were poured into an implementation project.

The result of the XITASO data quick checkconfirmed the initial hypothesis and showed that past requirements can indeed be used to calculate bid prices. This laid the foundation for further successful product development.

Similarity matrix of requirement texts

User Centered Design Process

Optimal usability through user centered approach

To ensure that the final software solution meets the concrete needs of the construction calculators, the UI/UX experts at XITASO focused on them from the very beginning. In addition to defining user groups and the persona “Edgar Effizient” as well as examining the user journey, click dummies were tested on real users. This enabled early feedback, on the basis of which the solution was continuously improved and expanded. Throughout the development process, this ensured that features and concepts were always implemented from the user perspective and that the usability and acceptance of the product on the market were ensured.

Always the highest quality in mind

High-performance solution thanks to vector database

In accordance with XITASO’s high quality standards, special emphasis was placed on performance and resilience in the digital solution for BRZ. Using language models based on neural networks, the semantics of the requirement texts were transformed into vectors.

The search algorithm now checks texts not for identical words but directly for the same meaning. Thanks to vectorization, millions of request texts are also searched in under a second and similar calculations are output.