Identifying Root Causes of Failures
To evaluate how well AI can classify cases into three levels – space, unit, and part of unit – in order to identify where the main sources of failures occur in the property portfolio.
Tid | 3 nov 2025 – 1 feb 2026 |
|---|---|
Plats | Digitally |
This AI project is co-funded by the EU – via the European Regional Development Fund (ERDF) and the Just Transition Fund.
What we want to solve
By analyzing the text in case descriptions, we want to automatically classify each case into three levels:
Space – e.g., kitchen, bathroom, hallway
Unit – e.g., refrigerator, stove, fan
Part of unit – e.g., door, handle, knob
This structure makes it possible to see exactly which spaces, units, or components generate the most failures – and why.
Example:
”We discovered that kitchen fans in a certain project generated the highest number of cases – which led us to change supplier.”
What value does it create?
With better insights into the sources of failures, we can make more accurate decisions about procurement, maintenance, and renovations.
Example:
”We decided to invest in a new type of refrigerator because one particular model had generated the most cases over time.”
What happens if we don’t do this?
We continue to purchase equipment that generates high maintenance costs.
We risk misallocating resources in our maintenance planning.
We miss the opportunity to reduce both case volume and costs.
For whom?
This test is aimed at those who:
Are responsible for procurement, maintenance, or renovation decisions.
Work with analysis or follow-up of case statistics.
Make decisions that affect operating costs and long-term property management.
The test group's mission
The beta program runs for about 12 weeks and is carried out together with a small group of selected real estate companies. The goal is to understand how the categorization works in practice and what value it can create in decision-making.
Registration, Form, and Selection (Weeks 42–44)
Fill in your name and email address above, and we’ll contact you with a short form to understand why this information matters to you. During this period, you’ll also manually classify 100 of your own cases. These will later be used to compare with the automated classification in Homepal.
Meeting 1: Precision and Classification (Weeks 45–46)
In the first meeting, we focus on how accurately the AI classifies your cases. Together we review the results and discuss:
Do the classifications match reality?
Is anything important missing, or does something feel unnecessary?
What kinds of decisions or insights could this help you make in everyday work?
We compare the automated classification with your manual assessments and talk about what works well and what might need adjustment. You’ll also see how Homepal categorizes your cases — and have the chance to influence how this should look going forward.
Further Development Based on Feedback (Weeks 47–48)
Homepal improves the classification based on what we learned in the first meeting. If needed, we’ll add new categories, remove unnecessary ones, or fine-tune the logic to make the results even more relevant.
Meeting 2: Using the Classification (Weeks 49–50)
Now we shift focus from testing to real use.
You’ll gain access to a dashboard showing your classified cases, where you can start analyzing:
Which types of cases dominate.
Where issues occur most frequently.
How the insights can support prioritization and planning of actions.
We also collect feedback — does this help you make faster and better decisions? After the meeting, we’ll make small adjustments to ensure everything works smoothly.
Beta in Use (Weeks 51–4)
During this period, you’ll use the dashboards and analyses in your daily work. Homepal follows up together with you, collects feedback, and makes small improvements along the way.
Evaluation and Closing (Week 5)
We end with a joint session focused on the most important question — what real value the AI classification has created in practice. The goal is to understand how Homepal’s solution needs to evolve to deliver even greater impact in real-world use, and to identify whether there are areas where the current setup doesn’t yet provide sufficient value.
Why participate?
As a participant in the test group you will:
Have a unique opportunity to influence the further development of the solution.
Gain early insights into which spaces, units, and components generate the most failures.
Take part in concrete discussions on how data can be turned into decisions that reduce costs and improve quality in property management.
Directly impact which features and priorities are developed next.
Participation requirements
The test group has a limited number of places. To participate, you need to:
Submit your role and a short motivation for why this information is important to you.
Describe which decision(s) you hope to make based on the material.
Commit to attending all three sessions and set aside time between meetings to review the sample cases.

