Starting up smoothly — Connecting government agency information through chatbots

In the chatbot network, in the same channel users receive answers from multiple organisations.

In public services we realise that customers often call the wrong organisations with their questions. Why does this happen? The distribution of responsibility is often arbitrary and changes over time. But what if customers don’t need to know which questions to ask from which government agency? What if there was one place where all information was united? And what if this place was made in a conversational way, where users can ask their specific questions and receive answers from whichever organisation this question belongs to? At Inland, this is what we work on under the project name of “Chatbot network”.

Chatbots to increase customer service quality

Chatbots are one of the hot topics to increase quality in customer service in public organisations in Finland. The idea of a chatbot is to automate customer conversations. This means the customers receive pre-written answers to their questions. When many public organisations simply do not have the staff to answer all client requests in good quality, introducing chatbots is less about replacing people by robots and more about increasing the quality of service.

From the customer point of view chatbots have several benefits:

  • They work around the clock, seven days a week.
  • They are as slow or fast as the customer is. There’s no pressure to understand everything correctly the first time. Users can read the answers several times.
  • There is no pressure to answer quickly. Customers do not feel that they are taking the time of other customers, so they can take their time to articulate their questions.
  • They allow customers to come back to a conversation after saving it.
  • They can be trained to speak more languages than a human customer service person could.
  • They are easier to navigate than a webpage, as there is no need to open many windows and search for the right information. Everything happens in the same window.
  • They mirror a natural conversation with a person. The user doesn’t have to learn anything new to interact with them.

Chatbot network concept

Since responsibility distribution is often illogical and changing, we at Inland are working on the idea to connect the chatbots that individual organisations in Finland are producing. With this idea we are bridging the gap between user needs and organisation’s siloed operations. Since this idea is appealing in January 2018 the Ministry of Finance has taken over the concept of developing what we call a “national network of chatbots” in Finland under the project name “AuroraAI”.

In the chatbot network each organisation owns an independent chatbot. These bots are connected via a “collective brain”, so they can forward customers to the organisation in charge.

There are two key elements of our chatbot network concept:

1. We organise the service around user needs, not government silos

As a basis for deciding on content we look at “events in life”. This means we create service-packages around user journeys that happen in specific moment’s in a person’s life. User needs in life-events always span several organisations, but when the chatbot network is in place the user does not need to know which government agencies are involved beforehand. Migri and its partner organisations are currently concentrating on the life-event “starting a company as a foreigner in Finland”.

2. Each organisation is in charge of their own content anyways

While we promote services around events in life which extend over government silos, we are also aware that practical operations of the agencies continue to be siloed. To avoid the well-known extensive collaboration projects between different public organisations with loads of planning, but little or late implementation, our chatbot network concept promotes each organisation’s responsibility over their own content area. The way we are technically building the content enables each organisation’s chatbot to work independently, but to transfer to all other organisations whenever needed.

Chatbot network demo and prototype

Since March 2018 Migri has experimented together with the Finnish Tax Administration (Vero) how the practical work behind a network of chatbots can be organised. We demoed a first version of a service for foreign entrepreneurs and workers wanting to come to Finland in June 2018. The main focus of the content was on advising the immigrants on applying for residence permits, as well as how to take care of their personal taxes. In early June we tested the prototype with several users. Some of them already lived in Finland, while others were in China considering starting a business in Finland.

In the demo event, civil servants could try out the chatbot network for the first time.

As an outcome of the first phase of the project we demoed the prototype in June 2018 to civil servants from different organisations in Finland. The recording from the event can be found here: https://youtu.be/YObCK8ITTDU?t=3283 (Some of the demo event was in Finnish. This video will start with the actual demo of the prototype).

Next step: chatbot network pilot

During summer 2018, a new organisation joined our collaboration: We welcomed the Finnish Patent and Registration Office. Now we extend the focus to questions about starting a company in Finland. Currently the project team concentrate on two company types: limited-liability companies or private traders. These are the three user personas we use to decide which content we need:

user personas used for prioritising content

From now, it is about one month before we plan to go live with our chatbot-network pilot for the first time. A lot of bureaucratic, team management and technical hurdles have been taken (finding rooms where 3 organisations can work together is still more challenging than you would expect) and we’re looking forward to testing our concept with real users. The pilot itself will last until June 2019.

Inland has had a big role in shaping this project: Our way of developing the chatbot network is implementing pilots with different public agencies, rather than building a huge infrastructure and hoping that everybody would join. In every phase of the collaboration we learn: How to collaborate with other organisation’s teams? Which content should we develop next based on customer needs?

We believe that this agile and collaborative way of working is the right path towards a smooth integration of chatbots in the public service.

So: Hold thumbs for our chatbot network pilot!

Edited by Mariana Salgado