How we accelerated tender processing with machine learning
Simplifying the complex process of (public) tenders in the HVAC construction space.
In the construction industry, tenders (invitations to tender for a project or part of a project) play a vital role in the procurement process for major construction projects.
Especially - but not exclusively - for projects from public institutions, complex documentation, inconsistent standards, and long waiting times for all parties involved characterize the tendering process.
Since both Viessmann and its partners (system installers) are constantly confronted with the issues of the tender process, wattx has been looking for a solution that would improve and ease the situation for these stakeholders.
✅ A machine learning powered platform that automatically matches relevant products to "Leistungsverzeichnis" positions.
After analysing the whole tender process, the required documents, involved stakeholders, and existing online platforms that are used for public tenders, we concluded that the greatest value for Viessmann in the current environment has to come from a faster and less manual handling of tender requests - specifically the process of analysing and editing the hundred-page long “Leistungsverzeichnisse”.
Due to the lack of standardisation in the document structure as well as the product descriptions, it is a task that has traditionally not been easy to automate. Based on our analysis, our long experience in applied machine learning, and recent advances in the technology made it now possible for a smart program to semi-automate this task. Consequently, we built a first prototype of the product to showcase its capabilities.
Viessmann was convinced of our initial analysis as well as our prototype, and gave us the greenlight to start developing an MVP that will enable the whole organization to improve their internal processes dramatically.
"Tender Service really showed wattx's expertise in breadth and depth for building an MVP.
Without the great tech team, applying advanced Machine Learning frameworks for Natural Language Processing would've been unfeasible."
Dennis Dümer
How we accelerated tender processing with machine learning
