Seamless Data Matching

AI-based automation of the complex tendering process in the heating, ventilation, and air conditioning construction sector.

CUSTOMER
Mappr
SECTOR

AI data solution

FOCUS
Data-based decisions
·
Automation
Mappr
Prozess

01

Research &
Interviews

We analysed market needs, customer problems and developed a business model.

02

Feature scope
& wireframes

Defining high-priority functions in collaboration with stakeholders and creating wireframes

03

MVP design
& development

Design of a user-friendly interface, and development of the backend.

04

Entwicklung & Markteinführung

Introduction of the MVPs with pilot customers whose feedback was used for product refinement and function validation.

ProCess

01

Recherche &
Business Model

Wir analysierten Marktbedürfnisse, Kundenprobleme und entwickelten ein Geschäftsmodell.

02

Feature Scope
& Wireframes

Definition von High-Priority Funktionen in Zusammenarbeit mit Stakeholdern und Erstellung von Wireframes

03

MVP Design
& Entwicklung

Design einer benutzerfreundlichen Oberfläche, und Entwicklung  des Backends.

04

Rollout
& Kundenfeedback

Einführung des MVPs mit Pilotkunden, deren Feedback zur Produktverfeinerung und Funktionsvalidierung genutzt wurde.

ProCESS

01

Initial research

User and stakeholder research through the use of personas, user journey maps, and interviews.

02

Business modeling

Definition of value creation, business model and planning.

03

Scope & prototyping

Defining the feature set for the MVP, specifying the functionality, and collecting feedback

04

Technical feasibility

Technical feasibility, cost assessment and implementation planning

PROCESS

01

User interviews

Conducting interviews with experts to identify pain points in manual workflows

02

Prototyping

Design of the first prototypes to validate the concept

03

Usability testing

Conducting interviews to collect feedback and identify issues

04

Implementation

Develop an MVP and publish it for the first user to collect feedback

Challenges

✕ Manual data reconciliation across different systems

Employees spent an excessive amount of time processing unstructured texts from complex documents, which were often up to 300 pages long.

✕ Common mistakes in data processing led to inefficiencies

High accuracy requirements under tight deadlines led to frequent errors and missed opportunities. Up to 30% of companies were disqualified from tenders due to errors.

✕ Fragmented and inconsistent data formats

Although GAEB formats are increasingly being used, many documents remain in inconsistent PDF formats, making manual allocation tedious.

✕ Scaling issues as data complexity grows

The lack of structured training data initially hindered the use of machine learning models.

AI-based automation of the complex tendering process in the heating, ventilation, and air conditioning construction sector.

Solution

✓ Automated data reconciliation

AI analyses tender documents to recommend items that meet the respective requirements, sorted by probability of relevance so that employees can make faster, informed decisions.

✓ Optimized workflow integration

High accuracy requirements under tight deadlines led to frequent errors and missed opportunities. Up to 30% of companies were disqualified from tenders due to errors.

✓ Increased accuracy and ability to learn

Mappr uses probabilistic matching for precise recommendations and adapts with reinforcement learning. User feedback trains ML models and improves future results.

✓ Adaptability for specific use cases

Tailored reconciliation criteria and workflows ensure flexibility and adjustment to specific business requirements.

TenderService has shown the breadth and depth of triebwerk.ai's expertise in building an MVP. Without the great tech team, the application of advanced machine learning frameworks for natural language processing would not have been feasible.

Dennis Dümer

Dennis Dümer

Head of Residential Pre-Sales and Quotation Center @ Viessmann Climate Solutions SE