This month, Capgemini is launching a new version of its omnichannel cloud platform, Odigo™. I interviewed Erwan Le Duff, Head of Prosodie-Capgemini, about the new platform features, the evolution of self-service, and the future of customer experience. Below is a video of highlights from the interview, followed by a brief introduction to Odigo and a full transcript of the interview.

Odigo at a glance

Odigo is a modular platform targeted towards customer care, sales, and service desk processes for end users, company employees, and citizens in the public sector. It has approximately 350 clients in 17 countries and handles over three billion interactions per year, primarily voice, but also chats, email, video chats, social media messages, SMS, and IoT transactions. It has four modules:

• Odigo Contact Center for omnichannel routing, recording, WFM, and quality monitoring
• Odigo Concierge, an omnichannel (voice and digital) bot, designed for self-service and qualification
• Odigo Campaign suite to push messages for marketing campaigns and notifications
• Odigo Analytics for customer insights.

Capgemini is releasing version 2.0 of Concierge with a chatbot developed completely in-house to enhance next-generation user qualification and self-service. Powered by NLP (Natural Language Processing), the chatbot processes customer questions and answers via texts, images, videos, documents, links, and mini applications, including choice buttons and selection carousels. It is offered as a web and app widget on a mobile SDK, and also on Facebook Messenger.

The machine learning algorithms at the core of the bot learn from interactions and prompt call backs, or schedule an appointment. In case of an unresolved issue, the bot can escalate to a shadow agent in real-time to provide the customer experience and train machine learning. Its API is open and the company can integrate it with existing databases, e.g. CRMs such as Salesforce, Microsoft Dynamics, Oracle, and Zendesk.

Capgemini has been implementing natural language-based IVRs for over 10 years and currently has 25 clients across sectors in English, French, Spanish, and other European languages.

What are the main differentiators of the platform and the new version?

Erwan Le Duff (ELD): It is highly scalable. This is one of the main differentiators, as we see today, more and more large-scale companies are looking for cloud solutions. They’ve been looking for CRM cloud solutions in the past, and are now looking at the cloud for interactions. We are addressing the needs of several thousand agent clients in a multi-country environment. We are also building the platform as open as possible to ease the integration of future channels, but also to help integration with the client IT systems.

We also provide end-to-end solutions. We can come to a client, analyze their current customer experience and current data, look at the customer journeys and transform them, measure the impact for the organization, and provide the technology in parallel to start automating, to deliver true omnichannel experience and break the silos that exist within all organizations.

Concierge v2.0 is really designed to address self-service needs, which are emerging everywhere, by introducing an omnichannel bot.

What drives these developments in self-service?

ELD: We see customers asking for self-service, obviously, but they are also asking for simplicity, quick responses, and 24/7 service. We see this emerging in all sectors. From the companies’ standpoint, there is clearly an interesting business case to automate part of their low-value requests. Depending on the client, we estimate that they represent between 15% and 30%. If they are able to automate part of those requests, it will free up time for agents to deliver more added value, outbound, or campaign management work.

Can you share more about the current implementations?

ELD: We started some months ago with a client in the insurance sector, where we looked at one of their service lines and reimplemented both a chatbot and a voicebot to manage the full process. A customer who wears glasses can call or chat with their insurance provider to know how much they will be reimbursed when they change their glasses. The system identifies the user, guides them through the process, including on the doctor’s paper prescription, and supplies the information. The bot then offers a short list of opticians based on geo-location data or pushes promotions for online sales.

Are you targeting any specific segments?

ELD: We see demand across sectors, but it is true that today some sectors are moving fast and need to transform the customer experience. For example, banks, insurance, retail, and finally, the public sector (we’re seeing this in few countries, including France). It is starting to become cross-sector and the reason is that the number of interactions is consistently increasing and clients are looking at potential automation as a kind of El Dorado.

However, what I say to clients every time is that before starting to look at automation and implementing bots we need to make sure we truly understand the call motives and make sure we focus on the right tasks. This is where our 10 years of experience in implementing the natural language part, including into the IVR, is a big help to understand the requests.

What prevents organizations from achieving omnichannel delivery?

ELD: Some of them are still organized in silos. The same people that manage the voice channel, don’t necessarily manage the email, chat, social media. We see this structure disappearing, but organizations need to fix it quickly. The second reason is the need to look at the agents and the agent experience, and have a strong vision and policy on how to organize channel management from the agent viewpoint. It is potentially a big change, and it may be easier when operations are outsourced.

Do you think the market has moved on from the initial chatbot hype?

ELD: We had a number of clients experimenting with automating certain processes because it frees resources and can increase customer satisfaction. However, it is not only a matter of technology. It is about understanding the call motives, implementing it on the right use cases, and more importantly, making sure the engine is strong enough and uses the right sectorial corpus to very quickly reach 95% natural language recognition. But organizations need to implement it in a true omnichannel situation. You can’t just have a standalone chatbot, because the customer may need to be transferred to a live agent or they may not like talking to a voicebot. It needs to be truly integrated into the omnichannel strategy.

Where do you see the balance between human-assisted and machine support?

ELD: People will be more used to speaking to Amazon Echo and Google Home, and this will increase adoption. Still, all companies know that the voice channel and talking to a human for certain inquiries is the only way to add sales. So, this element will stay, probably focused on more added value and more complex cases, but it will remain. We’ve seen in the last two years pure internet players with only digital channels now coming to voice. In four to five years the market can move to more automation, maybe even 50%, but not the majority of interactions.

Automation will potentially reduce the client’s need for agent support, but there is no way to stop this change. Most of the discussions we have with clients are not around removing 20% of the cost in the contact center, instead it’s about addressing increasing volumes and improving the customer experience. At Capgemini, we are proactive with clients to deliver a solution that will help them now and bring benefits in the future.

And finally... What are the next steps for the platform? What capabilities are you looking to add?

ELD: We started client tests with voicebots to finalize the omnichannel part the platform. It will be available in the next couple of months and will offer the same capabilities as the chatbot, based on the same foundations and data access. Our goal is to design the dialogues to adapt to the interface in a similar manner to the text bot, and at the same time, to fully integrate with Odigo modules.

We will continue to develop more connectors to third parties such as Salesforce. The last piece is to expand the use of AI and reinvest in NLP. For NLP, we want to further improve the solution to answer more complex inquiries, and we want AI to be used more and more for real-time, advanced routing of interactions, and ease agent experience by assisting them.