What is AI and how can it improve your customer service?


Johanna Mix, CEO of Barona Sweden interviewed Kirsi Louhelainen, who works as CTO for AI development, driving and overseeing AI work within Barona Group.  Below you can read more about Kirsi's insights on the things we are working on; for instance, recruitment matching algorithms, automated invoice handling, estimations, and optimizations for a number of problems from work shift management to dynamic pricing to finding patterns. AI and machine learning have tremendous potential in changing the work as we know it, and our job is to make sure it does so to the better. 


AI, or Artificial Intelligence, is generally defined as a computer being able to execute task at, close to, or above human level. It is not a single algorithm or technology, but rather a very broad and even ill-defined set of tools and solutions. These technologies vary from simple but limited rule based algorithms to today's hyped Deep Learning and statistical analysis which is very widely used today in applications from spam filtering to matching jobs to applicants.
The public image often sees AI as a general intelligence - a computer capable of solving a wide array of problems and even displaying creativity. However, such algorithms do not yet exist: all the cool things so far done with AI are actually about very specific applications. You can't take a model trained to win everyone in Go and make it drive a car. And you can't take an algorithm that is matching jobs to applicants and make it control a walking robot. In spite of fervent research in the domain, general AI is still far away.


AI is a broad concept and very versatile. It can be used in conjunction with different services to provide context and for example smarter and more accurate help. Technically, many personalization techniques used today in social media feeds, search engines and advertising are already utilizing some learning technologies. In a similar way, we can build algorithms that can guess what your issue might be if you are calling support, or find the right answer or product more effortlessly.

“If you already are using things like "old-fashioned" chatbots, you are well on your way. Some chatbots already provide some intelligence, others are more simple yet helpful."" 

This does not necessarily mean replacing people with software - it could mean that customer service people will have an assistant giving them crucial information and tips while they interact with the customer. Then again, repeated questions, getting initial information right etc. could be automated. Note that all automation does not mean it's using AI: you can have a surprisingly versatile helper chatbot with just prewritten answers and some manually written logic for selecting the correct one.


This depends whether you have designers, software developers, or data scientists in the house. Any of these can be outsourced, and definitely should if you do not have this skillset, but you also need to take clear responsibility of the development in-house. You need a person that stays on top of the project, communicates things internally, and makes sure the project responds to real needs.

Remember, though AI is much hyped and does require skillset that is at the moment hard to find due to demand, it's still all about process, software and general development. Any piece of software, however intelligent, is not going to magically solve your problems, unless you actually know what your problems are and what your customers need.

Any piece of software, however intelligent, is not going to magically solve your problems, unless you actually know what your problems are and what your customers need." 

Analyze your customer contact interfaces. How do people contact you? Do you have existing data that you could analyze for the types of questions and interaction you have with your customer? Do you already have the analysis made? If not, this might be a great place to start.


The bigger issue than a specific chatbot provider is to figure out how the chatbot or the phone service or the text notification or other facets of your customer interaction work in the big picture. If the chatbot does not direct customers to customer service when needed, it can be more of a hindrance than help when customer hits the wall and needs to reinitiate conversation through another channel. It might be more beneficial in some cases to bring AI to help customer support personnel in format of Intelligence Augmentation - providing help and cues during customer interaction.

“There's a wealth of chatbot providers out there, not all of which are AI powered though. GetJenny is one of the more known ones that also learns from your data, but the field is filled with fierce competitors from startups like ultimate.ai to big companies like Microsoft with their Bot Framework.

Also, it just might not make sense to automate a specific interaction - it might be much better to make sure that interaction never needs to happen. A case in point: if you go after your customer after they have decided to leave you, it most likely is a lost battle and no amount of automation is going to save you. Better to analyze early enough why the customer might be leaving and react early enough to keep them happy and loyal in the first place.

As said above, an isolated solution most likely won't make you happy but might rather be a form of trivial optimization. Working with a partner who can provide comprehensive service including process and general development can help you improve your customer experience and customer loyalty as a whole as opposed to optimizing specific parts of the process.

Thank you Kirsi!

Join the webinar for more interesting insights on the future of customer service!

New Call-to-action

Topics: ProductivityCustomerKundtjänstenAIFör företag