miva, occupation prediction – combining web interface with data from the physical world

Implemented owned by Rek.ai English
2y ago update
Read more at https://miva.se/
Miva is responsible for municipal water and sewage in Örnsköldsvik municipality. Miva is also responsible for collecting, treating, and transporting household waste.

They wanted a service-focused new web that wouldn't waste labor resources.

Limepark helped them build a new service web with AI integrated to show the most relevant e-services, the most relevant questions from the FAQ, and the innovative occupational prediction engine, which helps the visitor plan their trip to their recycling center.

Communications manager Lars Westberg likens the old website to an academic essay, with a summary, background, explanation, and hopefully a concrete guide at the end.

On the new web, the relationship is the opposite - we directly refer to the help and service available in the matter.

Read the extended version in Swedish on Limeparks website.

The objective was a higher service factor and, at the same time, reduce own administration.

The new website was also built to follow the new Accessibility Directive.

AI recommendations

Miva uses rek.ai's javascript to train their deep learning model, giving the editor the possibility to show real-time updated suggestions, sorted by relevance, instead of irrelevant static links that are chosen manually from time to time.

The new service web has AI integrated to show the most relevant e-services, the most relevant questions from the FAQ, and also a new feature;

Occupation prediction

Miva's website gives visitors an indication of how much congestion it is estimated to be at the recycling center. This is a solution that Limepark has developed for Miva together with rek.ai and is based on artificial intelligence. It's fully automated and delivers valuable insights for the visitor.

Miva's project contained a lot of new requests, which placed many new types of requirements on rek.ai. Data usually collected by visitors browsing a website would now be supplemented with data loaded from files. 

The deep learning model rek.ai also needed to be adapted to process occupancy. The result was a new type of AI model that enables projects that connect experiences from reality with better user experiences online.

Miva's newly launched website ranks 9th out of 443 tested public companies according to Webperf. This means that Miva's website proudly made it to the top tier for web-quality among all tested public companies.


Municipality, Region, Transportation & Logistics
Customer Service & Support, IT & Software
Better Customer Experience, Better Quality, More Efficient
Discovery, Prediction, Optimization
Sensor Data, Structured Data, Unstructured Data