Kellutin makes managing the Turku City library’s floating collection more controllable — and everyday work smoother

Turku City Library’s network includes the Main Library, 10 branch libraries, and 2 bookmobiles. Turku also serves as the central library for the Vaski consortium, which includes 18 municipalities and a shared collection. The volumes speak for themselves: Turku’s collection totals approximately 810,000 items, and in 2025 the library recorded around 3.4 million loans — and borrowing is trending upward.

Turku City Library needed an intelligent system to balance collections between libraries and make staff work easier. The goal was to route returned books to where they are most needed and free up more time for customer service. Kellutin was designed to meet this need.

“Our collaboration with Weasel Software has been close and has progressed smoothly under the guidance of a strong project manager! Overall, the change is significant, and after a year of use the results can be seen in better work rhythms both at the Main Library and at branch libraries,” says Kaisa Hypén, Service Manager at Turku City Library.

Over 20,000 books on the move every month

In daily operations, library materials are constantly on the move. Within Turku City Library’s internal logistics alone, an average of approximately 22,000 items of various materials are transported per month (Turku’s internal logistics). Of this, balancing accounts for an average of about 6,000 items per month.

Borrowing and return volumes — as well as demand for different types of materials — vary by branch. That makes the goal ambitious but clear: returned books need to be directed to the libraries where they are most in demand, while staff time is freed up for customer service.

Need: from the burden of “Teams requests” to a manageable process

Before Kellutin, balancing work between librarians was handled through various chat-based channels, most recently in Teams:

“Hirvensalo would need 15 generally interesting, fairly recent non-fiction titles”

“From Varissuo, Peppa Pig is completely out”

Of course, colleagues aimed to respond and help each other, but no one truly controlled the big picture. The way requests were made also made work fragmented: requests came throughout the day, interpretations varied, and the overall view was easily incomplete.

The goal was to build a system that is not just a channel for forwarding requests, but one that supports and develops the balancing process itself.

Solution: Kellutin — a learning collection-balancing tool to support librarians

Weasel Software built Kellutin for Turku City Library: an intelligent collection-balancing tool that suggests which materials should be requested for a branch and which should be sent onward.

“Already at the bidding stage, we received a solution from Weasel Software that was well thought out. We were truly positively surprised that an operator from outside the library field was able to understand our needs so well and deliver such high quality,” says Kaisa Hypén, Service Manager at Turku City Library.

How Kellutin works: data, algorithms, and user control

Kellutin is based on a nightly data extraction from Koha: the processing uses anonymised data related to items, titles, and borrowers. The algorithms calculate, for each title, the probability of being borrowed from each branch.

The calculation is influenced by factors such as a branch’s customer profile, borrowing profiles of different customer groups, overall demand for the title, and demand and fill rate within the relevant shelf section. The collection is divided into around one hundred “shelf sections/slots”, and each branch can adjust the target size for each section.

Users stay in control: Kellutin provides suggestions for materials to request or send onward, and suggestions can be filtered by location, language, material type, or classification. Kellutin provides up to 200 transfer suggestions matching the chosen criteria, and users can approve suggestions individually or as a “batch order”. A suggestion can also be rejected, in which case the item is placed in quarantine for a set period of time.

“An individual librarian typically thinks about the collection primarily at the level of their own library, and in that case there may even be a tendency to over-optimise the collection. Or you may wish you had those Finlandia Prize winners on your own shelf—hot titles in every library that get borrowed no matter where they are. Kellutin looks at the situation across the entire library network and makes assessments, for example based on borrowing data—something a human simply can’t do,” Hypén explains.

What changed: clearer rhythms, more precise requests, and fewer interruptions

In day-to-day work, the change is visible both for the ordering and the sending library. Balancing requests are defined in Kellutin using the specifications the system supports, making scheduling and organising work easier. When handling requests is no longer “dependent on a chat channel”, work can be consolidated and paced more systematically.

“Kellutin has progressed excellently from planning to deployment, and now to maintenance and further development. We have been satisfied, for example, with sticking to schedules and with the fact that project progress tracking has been understandable even for a non-technical person,” Hypén praises.

The solution was delivered as a customer-specific SaaS service from the Amazon Web Services cloud to ensure the service’s availability, flexibility, and security.