Chatbots have been around for more than a year now, and companies are starting to look at them quite closely. We’ve met with Clevy, a Parisian bot platform with a twist: they builds bots for corporate employees. What does that mean? It means they’re experts in making conversational interfaces out of HR procedures, helpdesks, change management processes or user manuals.
“Clevy was created during a hackathon in June 2016” said Bastien Botella, co-founder and COO of the startup. “It started as a side project but quickly turned into a full-blown startup.”
Why a focus on corporations’ employees?
“Well, companies everywhere are facing the same issues when it comes to internal communication or shared procedures. There are too many intranets, too many email exchanges, to many loose bits of information, all mixed in a huge yearly turnover. We wanted to find a way to centralise and facilitate the access to this kind of information for employees. By creating conversational agents that centralise IT procedures, employee on-boardings, days-off counters, and other workplace usecases, companies gain time, money, and peace of mind”. Your corporate chatbot is your new employee.
Why companies need bots?
“Bots are much more accepted in the corporate world than 6 months ago, let alone a year” said François Falala-Sechet, co-founder and CTO of Clevy “but even if our clients aren’t up to date with the concept, they quickly understand the benefits. When we explain that with our solution, they’ll get an interactive way to deliver selected information, available 24/7, on the communication channels their employees are already comfortable with, at any scale, they understand the power of chatbots.”
Do you think AI is essential for smart bots?
“AI is essential for bots if you want to provide an adaptable experience. Bots without AI can work fine, but reach their limits quickly: they can’t manage switches in topics, can’t adapt to the users’ way of speaking, cannot collect information in a non-survey manner. When you use Natural Language Processing, you bring modularity to your bot, and make it very scalable.”
“With SAP Conversational AI’s NLP and our own technologies, we have access to different kinds of metadata: sentiment analysis, sentence polarity, type and intonation. Because we can detect if the user is happy or angry, if his sentence is positive or negative, and if he’s asking a question or not, we can adapt our responses to provide the best service.”
What usecases do you find really clever?
“We’ve worked on a bot that started as a simple client-specific Human Resources FAQ, but evolved in a global “ask me anything about HR” agent, really valuable for employees.
Another project we loved was adapting Facebook’s Safety Check feature for Workplace. Facebook’s work social network doesn’t allow employers to access employees’ locations, something that is essential for the Safety Check. We therefore created a bot that reaches out to every workers in a company during an emergency asking “is everything ok?”, expecting for a yes or a no. That way, people can be marked safe without sharing private information.
Finally, Wysa.io is an emotionally sensitive chatbot that accompanies patients between their therapists appointments. It provides guidance, exercices and presence to people in need, and is overall really cool.”
3 tips to make an awesome bot?
“First, start slow. Bots aren’t magic, they’re logic. Start with an easy use case, where a bot does one major thing, and does it well. Then add another feature, and keep building.
Second, do not underestimate small talk. Most people expect bots to be clever and witty. If the user says “I love you” and the bot replies with “Sorry, I didn’t get that”, they’re disappointed. Create extensive support for basic chitchat. “
"Bots aren’t like websites, they’re not fixed, they evolve with time. "
“Third, monitor, monitor, monitor! Bots are always in motion, and can always better adapt to market needs. Conversational interfaces have one great strength: people, through no intermediary, are telling you what they want! So take it, implement it, and keep improving your bot until it gets it right every single time.
And if you need any help, give us a call 😉 “
In the following months, Clevy wants to make bot building even easier. They’ve already reduced bot making from months to days, and now want to make it a matter of minutes!
Thanks to Bastien Botella and Francois Falala-Sechet for their time. You can discover Clevy here, contact them to build a bot here or apply for a job (they’re hiring!)