Koïos Intelligence has just launched Olivo, a specialized, insurance customer-focused, cloud-based application using natural language processing (NLP). The new technology has attracted major investors, securing $6.5-million in funding in a recent financing round.
Koïos Intelligence is a Quebec-based company founded in 2017 by Mohamed Hanini, its CEO and chief technologist. With a bachelor's degree in actuarial science and a master's degree in mathematical and computational finance from the University of Montreal, he launched Koïos Intelligence with the general goal of finding solutions and improving processes, specifically in the insurance domain in life and health and property and casualty.
"We are thrilled to support and welcome a visionary company like Koïos into our startup ecosystem, offering an innovative solution that enhances the experience of all parties involved throughout the value chain," expressed Geneviève Biron, founder and president of Propulia Capital, an investment platform that supports technology companies. Brion recently joined Koïos' board of directors.
A look at Olivo
Olivo directly interacts with clients through online chat, provides a wide variety of insurance product information and the firm claims a seamless and efficient claims processes.
"The objective was to reduce waiting time and offer a new interaction experience for consumers, not only for purchases but also for updating client information," the firm’s founder says.
"The main idea is to process real-time information and make life easier for consumers while facilitating the work of brokers. Brokers will evolve towards providing more personalized advice," he added.
The conversational agent then becomes a virtual assistant for the broker, allowing quick access to consumer research and contract updates. Moreover, the user always has the option to speak with a human agent during their conversation with Olivo. The application follows a 100 per cent hybrid approach, providing consumers with the choice to interact with both artificial intelligence and a broker, Hanini explains.
Specialized application in its field
The Olivo agent has been trained to analyze and classify user requests. It employs multimodal artificial intelligence (AI), a type of AI that utilizes multiple channels of information transmission through algorithms. If a consumer asks questions unrelated to insurance, the robot will be unable to respond. The application employs a specific deep learning model, established within a neural network, to predict the user's meaning and intentions based on structured data.
"After identifying the user's intention and understanding the instruction, the machine needs to predict an action and consequently provide a response to the question," explained Hanini. Olivo has been trained to give precise, prepared, and controlled answers based on its specialized insurance database. This distinguishes it from typical conversational agents that generate responses by relying on information found on the internet.
Hanini tested Chat GPT by asking whether the color of a car could influence insurance premiums. He says the Olivo conversational agent has been trained not to make mistakes when faced with such questions.
Beyond online chat, Olivo possesses a revolutionary feature in the realm of conversational agents, according to Hanini. Users have the option to address the robot vocally – the agent has been trained to receive instructions or questions through voice recognition. In addition to this capability, it can communicate in forty languages.
Confidentiality and security
Ensuring data protection and securing users' personal information is essential for the company. Koïos Intelligence has enlisted professionals specialized in cybersecurity, risk management and operational chains. It has also partnered with various experts from the international scientific community, including the Institute for Data Valorization (IVADO) and the Montreal Institute for Learning Algorithms (MILA).
Another key aspect of confidentiality is that the Olivo agent has not been trained to reuse user's personal information but only to store it directly in the client's database.
"In our development environment, we do not have access to the user's identity. The model does not consider this information. Moreover, the robot has not been trained to retain user's personal data. The conversational agent's neurons have only been trained with insurance-related data from our specialized database. They are primarily prediction models," Hanini adds.
The company says it is committed to the responsibilities and obligations it has regarding clients. Consumer data is the property of the client – it’s a clause stated in their contract. "Apart from the broker or insurer, no one else can use the consumer's personal data. The information is protected and controlled by the client," Hanini adds.
He also emphasized another important point regarding server geolocation. All of the company’s servers are based in Canada and are therefore governed by the country's laws, which protect personal information in the private sector.
Other projects
The company is continuously improving the performance and expanding the functionality of its Olivo application, according to Hanini. Currently, they plan to establish a "modern" call center managed by artificial intelligence. Users will be able to interact with the conversational agent via phone, utilizing voice recognition. Conversations will be recorded, allowing the agent or the broker to access information already expressed by the consumer in real-time.
Koïos Intelligence is also working on a second funding campaign, scheduled to take place in a few months. With locations in Canada, the United States, and North Africa, the company has also recently expanded its presence to France.
The company relies on demonstrations through its website to increase its visibility and awareness.
