Case study

How the Finnish social insurance institution consolidated 8 chatbots into 1 virtual agent on the platform

Kela - Finnish social insurance institution

Kansaneläkelaitos (or Kela, for short) is the Finnish social insurance institution, a government agency in charge of settling benefits under the country’s national social security programs. Originally founded in 1937 to handle retirement pay, today it encompasses a variety of social programs including child benefits, unemployment benefits, sickness benefits, health insurance and student financial aid.

Kela's results at a glance...

FAST: Virtual agent built and launched in under 4 months
EASY: Zero developers needed to migrate 1,500+ intents during VA build
SCALABLE: 8 basic chatbots consolidated into 1 advanced virtual agent

In 2020, Kela reported that its online services were accessed more than 64 million times, with 72% of all applications processed via digital channels. This shift towards digital in the Finnish population is also reflected in Kela’s approach to customer service, with the agency choosing to supplement its existing support channels with a chatbot solution as early as 2017.

However, this previous chatbot platform came with a number of constraints. Kela was limited to creating eight separate chatbots for each of its benefit programs, making it challenging to maintain and optimize in order to keep content fresh and consistent for users. 

From chatbot to conversational AI in under four months

Kela had outgrown the limitations of its current solution and, in 2021, held an RFP in which’s conversational AI platform was selected as the winner. Together with implementation partner Front AI, Kela leveraged’s advanced Natural Language Understanding and user-friendly conversation builder to develop Kela-Kelpo, a virtual agent powered by conversational AI with capabilities that far exceeded the previous chatbot solution.

In less than four months, Kela was able to transfer all of the data from the previous system into the platform. Over 1,500 intents (or topics) from eight different chatbots were consolidated into a single virtual agent designed to answer questions in both Finnish and Swedish. Launched in March 2021, Kela-Kelpo’s initial scope encompasses frequently asked questions related to family and child benefits, as well as income support, with plans to extend its knowledge base to cover student benefits-related content in the near future.

Kela selected because of its superior technology which allowed for more functional and efficient communication between end-users and the virtual agent. Using deep learning and natural language processing, Kela-Kelpo is able to contextualize answers based on whichever page of the agency’s website the user is currently on. i.e. if the user is on a page about child support, the virtual agent can answer questions related to this topic without the need for the user to mention child support in their query.

Using’s no-code conversation builder, Kela’s team can continue to train and optimize the virtual agent over time, analyzing conversation data in real-time and easily adding more content to cover new customer inquiries. This is all managed without any previous technical experience and in a single environment so that updating is quick and easy - a far cry from having to update eight chatbots individually, saving the agency time and resources.

A pioneer in the early adoption of chatbot technology, Kela’s migration from a first-generation chatbot solution to’s advanced conversational AI platform is a testament to the agency’s willingness to embrace innovation and its drive to continue delivering the best possible automated customer experiences.