NL | FR | EN

AI, Artificial Intelligence, in customer contact


IPG, member of Koramic2Engage, is strongly committed to the ongoing development of innovation, talent and technology. IPG therefore sees opportunities for establishing strategic partnerships with companies that want to grow together. These are organisations that, like IPG, make a priority of operational excellence and customer intimacy and that enable IPG to further develop the talents of its own employees. IPG recently set up a partnership with Trendskout. This partner offers a platform for artificial intelligence (AI) which allows organisations to create and automate further AI applications at their own initiative. In this interview with IPG Chief Operations Officer, Wim Hofmans, IPG IT Development Manager, Danny Van Woensel & Co-founder of Trendskout, Pauwel Grepdon, you’ll learn how this collaboration came about, the steps involved in the AI implementation process and what else the future of artificial intelligence has in store for us.

Wim H. & Danny V.: In a world where digital transformation, first time right and time 2 market have become indispensable, IPG considers it necessary to be just as agile as an organisation, to continuously optimise operations, to innovate and to further develop relationships with our partners. An important element remains our human capital; our employees, whom we support in their development from experts to generalist experts. This means that we develop thorough knowledge in one area, but also provide the basic knowledge of several other domains within customer interactions. As a result, our collaboration is even stronger, our organisation is more resilient and it can move forward with an even greater sense of autonomy and ownership.

What’s more, instead of getting lost in day-to-day operations, IPG is creating more space today to make innovation second nature. A data-driven organisation is key here, and a Data and Reporting Office has been created for this purpose. IPG is also looking for strategic partnerships that support its vision and mission. Among the consultancy options, Trendskout stands out because they offer an AI platform that fully fits in with the strategy that IPG pursues.

Pauwel G.: Trendskout does indeed approach Artificial Intelligence in a different way. We work from a product approach rather than a project approach. For example, partners, such as IPG, can use our platform to develop AI cases. Trendskout includes several immediately deployable AI solutions ranging from Speech-to-Text to text classification based on Natural Language Processing. In addition, all models can be linked via a flexible input and output. Speed is a big plus here, because we don’t have to set up a new project for every new AI-related topic. The collaboration with IPG has grown organically. The preparatory discussions crystallized into a number of concrete cases.


Pauwel G.: Within Artificial Intelligence, text and speech processing was already one of Trendskout’s pillars. So we’re pleased to be able to work with IPG. The close cooperation with IPG and in turn, with its partners also means the chance to build up even more expertise in that area and that we will be able to significantly expand the possibilities.

There’s a lot of potential in this collaboration for us as well as for our own partners.

Wim H.: We also want to further consolidate this collaboration in order to have more impact together, in a supportive way. There’s a lot of potential in this collaboration for us as an organisation as well as for our own partners that we want to make the most of.

Wim Hofmans – Chief Operations Officer at IPG (AI)
Wim Hofmans – Chief Operations Officer at IPG (AI)


Wim H & Danny V.: First of all, we drew up an internal long & short list of cases that we want to realise at IPG using automated AI. Several discussions were then held with Trendskout about this and the choice was made to start with a first case that would immediately yield successful results. This greatly strengthened the support for automated AI within our organisation.

However, there are 2 important conditions that have to be in place to set up a successful case, including, firstly, quality input. Quality historical data must be available that can be exchanged. Secondly categorisation should also be possible. This is a way to qualify the data/input. It is this data and categorisation that is then picked up by Trendskout in a subsequent step. Examples here are voice logs along with the manual action of the contact center professional.The AI algorithm is then allowed to process this input in order to compare the outcome generated by the AI with the historical data of the own employees.

It’s important to note that we do not work with a naive model. After all, a naive model would go through the data and always favour the classification that occurs the most frequently. The analysis model we use takes the classification imbalance into account and also looks at what the choice should be if the most common class is filtered out. In order to allow the algorithm to achieve an ever-increasing success rate, Reinforced Learning is used. This makes it possible to have given outcome with a given degree of certainty from the model checked by employees and to provide this data again as feedback to the algorithm. In this way, the algorithm receives further confirmation and continues to learn.

We can consider a model to be of high-quality if it is on average 70-75% likely to classify correctly. Through reinforced learning we can increase that probability to 80 to 84%. A final element that should not be underestimated is the willingness of the partner to grant us access to make links in our own systems, in order to be able to use the data. This is not always an easy step, but it’s indispensable in such processes and absolutely crucial for validating the hoped-for efficiency via AI. GDPR compliance, Secure Data Exchanges & Data Retention Periods are also important elements here.


Danny V.: I wouldn’t really say barriers. It is certainly the case that the entire process is relatively time-consuming. In our day-to-day business, we are often asked to act fairly quickly. However, the model also needs time to learn through data exchange or reinforced learning. Training a model takes an average of 3 weeks, depending on the complexity. Of course, it’s not only the technology, but also the employees who are included in this new way of working so that that they also know how AI collaborates with them.

Pauwel G: Setting up a new AI implementation is a process of learning moments. However, you should always have a starting point that will deliver the necessary benefits. If an initial model is already 70% accurate, you can use it to automate a substantial part. This way, you reap the benefits in the short term and you can then fine-tune it further.

Pauwel Grepdon – Co-founder Trendskout (AI)
Pauwel Grepdon – Co-founder Trendskout (AI)

Setting up a new AI implementation is a process of learning moments.


PAUWEL G.: An important trend we notice is Conversational Artificial Intelligence, where the system comes into direct contact with the customer. Today, AI is often used to interpret the question, but this is still linked to a human action. The next step in this concept is that artificial intelligence itself will also be able to answer a customer’s question via text or speech.


Wim H. & Danny V. & Pauwel G.: Specifically for managing customer interactions, we see a lot of possibilities. For example, we see options to:

– manage customer interactions even more efficiently and effectively through data mining and Speech-to-Text
– automatically make standard requests accessible by converting Text-to-Speech to data
– emotion detection to give customer service agents a possible direction for the conversation with the customer
– targeted guidance of employees in their development through data
– transforming the IVR selection menus into a system where the end user only has to ask his question and is then automatically forwarded to the correct service or where AI automatically provides the solution

These are all cool and exciting developments.

Of course, it’s not only the technology, but also the employees who are included in this new way of working so
that that they also know how AI collaborates with them.

A question we often get is whether AI or robots are going to replace humans. We absolutely don’t believe that. We believe in a hybrid model in which Artificial Intelligence is at the service of our employees. As a result, the repetitive standard actions can be handled by AI and the employee can focus on the valuable and knowledge-intensive interactions. The advantage of IPG is that in addition to the technology, we can also add the human emotion of our highly customer-friendly employees. By working in this hybrid way, we can ensure that we will be able to continue to create the best possible conditions for our employees and continue to build on this.


Wim H.: It’s true, we’re not just talking about concepts but we’re already converting them into concrete implementations. In addition to continuously working on the short & long list as we mentioned, we already have an AI application live today that categorises voicemail messages through automated AI without human involvement. The intended goal: TCO reduction and internally to take a first concrete step towards AI in order to create an even broader support base.

We will also share concrete business cases through our channels in the coming months. So, keep an eye on our social media and for more news about customer interaction management, concrete business cases and developments in the industry, or contact us to discuss the possibilities for your customer contact.

Want to know more about how Belgium’s leading contact centre can strengthen your customer contact with AI applications? Contact us.


Leave a Reply

Your email address will not be published.

About IPG

IPG, part of the Koramic Investment Group, is a European call center group with subsidiaries in Belgium, the Netherlands and Morocco. 3.000 staff members guarantee a professional handling of all your customer contacts, 24 hours a day, 7 days a week. In Belgium, IPG owns the largest call center, with over eighteen years of experience.

Therefore, we can rightfully call ourselves expert in customer interaction management. We are your business partner by excellence to manage and optimise the most important pillar of your business. We are well aware that your customers are the driving force of your business and precisely because of that reason we want to offer them an optimal customer experience.

With the rise of the social media and unlimited access to the Internet, your customer service has become more than ever the touchstone of your marketing and sales departments. Your corporate image depends on what is told about your organization, your products and your services.

Our people are qualified to confirm your brand identity or lead your existing and future customers into the direction of your products and services.

Finally, new acquisitions, further internationalization and a continuous focus on innovation enable us to look to the future with great confidence.