All entrepreneurs know that first of all they must make a deep analysis of markets and then offer needful products or services. One of the increasingly popular ways in the business to business field is direct collaboration between startups and corporations when one side says their needs and another side creates a needful solution.
Past three years the international project Scale-up Champions has been a great stage for deep-tech startups to find new clients and partners. Mateusz Jarus, the founder of the Polish startup Expansio Software House, is glad that his team had the opportunity to participate in the Startups and Corporates Collaboration Programme and cooperate with the company Volkswagen Poznań.
The app was downloaded by 5000 employees in 3 days
Mateusz Jarus reveals that the corporation had the clear challenge - internal communication with employees. This issue became very huge in the period of pandemic when all employees must stay at home and wait for needful information.
According to Mateusz Jarus, they created the mobile application for internal communication. This app is very similar to modern social media platforms, like Facebook, and it is so easy to install
“This solution was quickly accepted by the company. When the pandemic started and the production was closed for a few days, the application became the first source of information for employees who needed to stay at home. In just three days it was downloaded by over 5000 people”, says co-founder of the startup.
This technology is comfortable to use because employees do not have access to any business phones. They needed an efficient and quick solution for their internal communication.
Solved the problem of internal communication
Mateusz Jarus says that Volkswagen asked if it would be possible to just upload documents to the chatbot so that he could read them automatically and answer questions based on the knowledge inside these documents. That was the beginning of BigData Chatbot.
According to him, it quickly turned out that people often called the hotline with many questions about using the application or about their daily work. Volkswagen asked for a chatbot implementation that could automatically answer any questions. However, chatbots have one major flaw. You need to create the set of questions and answers manually.
“We applied for a grant from the National Center for Research and Development. Our idea was to implement natural language processing algorithms that would work similarly to Google. When you ask a question to Google, you often get not just the list of websites but also the exact answer to your question in the form of a few sentences. This answer is retrieved from one of the websites. We do a similar thing but we focus on the documents and the algorithms work inside the company. You may send thousands of documents to BigData Chatbot, ask a question and get the exact answer. Our project simplifies searching through large amounts of information", says Mateusz Jarus.
According to him, today BigData Chatbot works with Polish corporations that help build the tool. These companies saw the potential of the tool and decided to give the startup their internal documents to build the training set for machine learning algorithms.
Ready to adapt their solution for free
Mateusz Jarus says that Scale-up Champions was the opportunity to find international partners.
“It was a long process. It is not easy to convince a large organization to actively cooperate with a small startup and even more difficult to get any documents from them. However, we found companies that see this tool as an opportunity to solve their internal problems. One of them is a bank, where the security guidelines are very strict. It took a few months to finally receive some documents”, says Mateusz Jarus.
According to Mateusz Jarus, these documents are very valuable for BigData Chatbot. Machine learning algorithms need the so-called training set to learn their job. Here their job is to answer questions. In order to be able to answer any question, they need a lot of examples.
It is a very time-consuming process, but a crucial one. The quality of these documents, questions and answers will influence the quality of the final algorithms.
“We are looking for large organizations in any sector - automotive, banking, finance and more - willing to improve the management of the information. Our project shortens the time of searching for an answer. Since we are financed from a government grant, we are ready to start collaboration for free. We can prepare the machine learning model for the company, deploy it and prove its value in the selected department. In return we would like to receive some documents used inside the company to train the models”, says Mateusz Jarus.
He says that it is hard to gain the trust of a foreign company and start cooperation, even for free. Scale-up Champions gives the opportunity to present the project in front of a wider audience and reach new contacts.