Artificial intelligence programs used to consult the public in Lithuania

Brief success story

In the first five months, Simas, a virtual assistant at the State Tax Inspectorate, responded to more than 94,000 inquiries on issues relating to land taxes, real estate taxes, individual activity and business licenses, payment of fines and subsidies to companies. Working around the clock, the assistant communicated with and provided advice to more than 33,000 consumers.

Not only Lithuanian businesses, but also state-owned enterprises and public service providers are moving into the digital space. They rank sixth in Europe according to the Digitized Public Services Index, which is above the EU average. An increasing number of public sector companies in Lithuania use digital solutions based on artificial intelligence technology. One such solution is Simas, a virtual assistant advising customers at the State Tax Inspectorate.

Understanding customer habits is essential for the training of the virtual assistant

The virtual assistant of the STI, launched back in December, provides residents with general information on land tax, real estate tax, individual activity and business certificates, payment of fines and subsidies to companies. It helps taxpayers get general information faster and easier around the clock, without lunch breaks and without waiting in line for a telephone consultation.

“By creating a virtual assistant, we sought to improve the quality of advising residents, which is especially important now, during quarantine. It should help citizens get access to general information more quickly and easily, without limiting the number of inquirers.”

Lina Kurlenskaitė,
adviser at the STI Tax Information Department

“We hope that this will help reduce the number of people waiting for a telephone consultation,” says Lina Kurlenskaitė,

According to Renata Špukienė, director of the language technology company Tilde Informacinės Technologijos, which implemented the innovation, teaching a virtual assistant to properly answer all the questions received from customers is not an easy task and requires constant data analysis.

“In this case, the interview robot trainers, who teach the STI’s virtual assistant Simas to answer questions, must understand customer habits, predict possible question options and ways to ask them, understand the sequence and logic of questions, and at some point even become demanding customers themselves.”

Renata Špukienė
director of the language technology company Tilde Informacinės Technologijos

“All of this requires both a lot of time and effort. Only then can the final result meet customer expectations and the customers will have an easy time ‘talking’ to the virtual assistant,” says R. Špukienė.

Challenges in developing artificial intelligence solutions

This is also supported by L. Kurlenskaitė. According to the STI specialist, the use of a virtual assistant requires constant improvement and training of this tool.

“In our case, most topics have similar keywords, so the biggest challenge is to teach Simas to differentiate, for example, real estate tax from income tax on property sold, and not to get lost between the different tax returns. Simas is capable of knowing everything about taxes, but with each new topic, his teaching also becomes more complicated,” says L. Kurlenskaitė.

Virtual assistant reduces service time to 5 minutes

According to R. Špukienė, a virtual assistant significantly reduces the time spent by employees answering basic questions received from users. As a result, more human resources can be allocated to more complex problems and issues. A properly trained virtual assistant can save you up to 30% of the time spent serving customers, and in some cases even more. However, in order for the general public to have convenient and fast virtual assistant services, they must first be taught to public sector employees who work with them.

“Public sector employees deal with both simple inquiries and complex issues of residents or businesses by providing direct advice or telephone consultations. Taking the time to solve the simplest questions leaves complex issues that require human intelligent unanswered in time.”

Renata Špukienė
director of the language technology company Tilde Informacinės Technologijos

“In solving this problem, we aim to demonstrate how artificial intelligence can accelerate and improve the provision of public sector services and the satisfaction of residents and businesses,” says R. Špukienė.

AI will provide at least 30% of customer services in the future

Users are already looking for answers to their questions through all possible channels, including the internet, social networks, their phones and apps. Therefore, organizations can no longer rely solely on the help of customer service professionals and must use the omni-channel service principle.

“Almost half of Lithuanian companies share information by electronic means; residents willingly engage in various activities on the internet, providing an ideal starting point for the implementation of digital services in the public sector.”

Renata Špukienė
director of the language technology company Tilde Informacinės Technologijos