of user’s request will be managed by the chatbot in 2020
understanding rate with Konverso AI conversational platform
employees coveredby the chatbot
With the ongoing digital transformation, end users’ habits and practices are changing. Business users ask to work from everywhere. They need to collaborate with colleagues and partners all around the world. They want to be more efficient and productive in their daily work. To meet their employees’ expectations, companies need modern tools and collaborative solutions, both secure and flexible. They also need to ensure that these solutions are widely adopted.
To help its customers increase user adoption and engagement, Computacenter launched a digital workplace strategy, called Digital Me. DigitalMe is a workplace designed for people, engineered for business. It helps CIO accelerate their digital agendas with agile workplace solutions that unleash the power of their people and enable business success. It equips, empowers, and assists people with solutions that are increasingly powered by analytics, AI and automation to reduce costs and provide a proactive digital experience. Delivering intelligent support aligned to personal preferences is a key component of this strategy.
“Our goal is to enhance agility, make end users more autonomous and allow real business successes”
According to Arnaud Lépinois, Managing Director of Computacenter France, DigitalME addresses three challenges:
“Our goal is to enhance agility, make end users more autonomous and allow real business successes,” says Arnaud Lépinois.
Designed for faster resolution of people’s technology problem, AssistMe leverages Computacenter AI & automation capabilities to be able to anticipate and fix issues before they become impactful to the user. It offers a choice of contemporary solutions – social, mobile, chat, face to face – that allow people to consume support in a way that is convenient to them. Access to solutions and the right support experts when needed for IT, and increasingly more so, non-IT issues. Underpinning these solution is Computacenter extensible digital delivery platform called AIMY, which is based on three components: artificial intelligence (AI), automation, and analytics.
Computacenter wanted to bring the capabilities of this platform directly to the end users, and to do so added Chatbot into the End User engagement channel mix to address issues automatically and provide intelligent knowledge and assistance without the need to connect with a human agent.
At a Microsoft event, Computacenter discovered Konverso conversational AI platform and virtual assistant solution. “Konverso approach and technology matched with our expectations, so we decided to integrate it with our platform,” remembers Arnaud Lépinois.
“A standalone Chatbot doesn’t bring a lot of value: the bot needs to be part of an ecosystem in order to deliver real business benefits”
“Before proposing a solution to our customers, we deploy it internally,” states the MD. During the summer of 2018, Computacenter started to implement the Chatbot for its employees. Computacenter’s maturity regarding employee support is higher than most organizations, where self-service contributes to a high proportion of contacts made: many simple requests and issues are self-resolved via an extensive knowledge base and automation through a support portal, allowing the Service Desk to handle more complex business issues. To allow a chatbot to embed itself into this established engagement platform, Computacenter’s Product Manager for AssistME, Alan Connolly, had strong requirements:
“A standalone Chatbot doesn’t bring a lot of value: the bot needs to be part of an ecosystem in order to deliver real business benefits,” warns Alan Connolly.
The project team started by analyzing the processes that they planned to automate. “First, we wanted to script them, but soon, we realized that it wasn’t possible: service desk tickets encompass more than 8,000 use cases inside our company,” explains Alan Connolly . Then, the team quickly shifted its strategy: they decided to use Konverso AI capabilities of its platform instead of coding every use case manually. “We opened our knowledge base to Konverso solution, to let the bot learn by itself. We also gave it access to one year of users’ interactions with the service desk.” After this step, the bot accuracy took the leap: its natural language understanding rate reached 80% “even before it became available to end users,” says Alan Connolly the Product Manager.
During the project, the company encountered a few other challenges. One of them was related to the data used by the machine-learning algorithms, especially users’ data. Once personal data is involved, its use is subject to regulatory agreement. To comply with GDPR, data protection officers were consulted in every location across the Group. In some cases, data had to be anonymized before the Virtual Agent could use it.
In June 2019, Computacenter started to roll out the solution, country by country. Today, the solution is deployed in three languages (English, French and German) and covers all 15,000 employees. “We already noticed a rise in interactions. Just a few months after the launch; the Chatbot already manages 17% of users’ requests, which is high regarding our end users’ maturity,” states Alan. Computacenter expects to handle 30% of user requests with the Chatbot by the end of 2019, with 20% fully automated, and 50% by the end of 2020, with 40% automated scenarios.
“We already noticed a rise in interactions. Just a few months after the launch; the Chatbot already manages 17% of users’ requests, which is high regarding our end users’ maturity”
For service desk agents, the project is an opportunity. As the Virtual Assistant frees them from repetitive tasks, they can focus on more complex requests. Computacenter is currently developing their business skills, allowing the agents to fulfill new value-added missions.
“On this kind of project, the average return on investment (ROI) is 24 months,” says the MD. The higher savings are observed once some level 2 and level 3 requests are automated, because the cost of a ticket grows exponentially with the complexity.
The project also delivered non-financial benefits, such as an increased level of satisfaction among the users of the service desk. “The Chatbot can also be used to enhance the adoption of new solutions, such as Microsoft Teams, or to support business processes, such as HR and Facilities solutions,” adds Alan Connolly “The true ROI of the solution lies there, in complex business use cases.”
To fulfill tasks such as solving the end user’s issues, a conversational AI platform must be tightly integrated with the information system. “Replacing the search engine of a knowledge database by a Chatbot isn’t enough to deliver real benefits,” says Alan Connolly. The Virtual Assistant should be able to interact with various internal applications, such as Single-Sign On (SSO) systems, RPA, ITSM, HR management software and even other bots. It should also be able to use external services, such as real-time translation tools or image recognition. “Konverso provides a set of APIs. This openness allowed us to connect the Virtual Assistant with our ITSM solutions (BMC and Servicenow), our RPA suite (Blue Prism) and Office 365. We also plan to integrate the Chatbot with our Success Factors suite, in order to automate actions such as vacation scheduling,” describes Alan Connolly.
Another key lesson is the strategic place of human expertise in AI-related projects. “Data scientists are key to give a meaning to raw data, as well as the understanding of users behaviors,” emphasizes Alan Connolly, who adds, “IA isn’t magical: it only learns what you taught it.”
“We have chosen to invest both in people and in automation: this allows us to be very competitive compared to offshore service providers”
The last lesson is that a Chatbot project must be managed like a transformation program. To succeed, the project has to follow a given trajectory: the technical foundations, the data and the human expertise must all be in place before starting to deploy the bot. “Sponsorship at the executive level is also a key requirement: the program leader has to manage the project in all its dimensions, from security to compliance, including accurate communication to encourage adoption”.
AI is often feared because of its potential social impact. In this project, AI implementation didn’t lead to people replacement. Instead, Computacenter plans to recruit more of them, because the company has gained more customers with its unique approach. “We have chosen to invest both in people and in automation: this allows us to be very competitive compared to offshore service providers,” concludes Arnaud Lépinois.