Bondzai: Industrial AI That Learns from Humans
Born out of cutting-edge university research straight from the Alexander Grothendieck Institute (UM) and an entrepreneurial approach focused on innovation, the startup Bondzai has developed an artificial intelligence system capable of learning in real time and continuously within a complex industrial environment. It is currently preparing an ambitious fundraising campaign to raise two million euros. Interview with its co-founder and CEO, Alain Fanet.
What is the nature of the relationship between the startup Bondzai and the University of Montpellier?
It all began with the work of Bijan Mohammadi, a mathematics professor at the University of Montpellier and a researcher at the Alexander Grothendieck Institute. In 2016, he wondered how to use artificial intelligence to simplify mathematical problems associated with numerical simulation. He drew on his 30 years of experience in this field to automate and optimize AI learning processes, leading in 2018 to a new learning algorithm, which he shared with Satt AxLR.
What led you to co-found a company based on this academic research?
I was looking for an AI project that broke away from the mainstream to make artificial intelligence more accessible and simpler. In late 2018, Satt AxLR told me about Bijan Mohammadi’s project, and in early 2019, they proposed a one-year pre-maturation program to assess our ability to work together. We come from two very different worlds: academia and the private sector. In late 2019, we launched a maturation project with a budget of nearly one million euros. And then the health crisis hit. On December 28, 2021, we finally founded the company Bondzai with all six people who remain shareholders today. Bijan Mohammadi also participated in the creation of this startup through a science competition (watch Décollage#3 : The Science Competition). We had an algorithm and initial voice-centric prototypes—specifically voice control—targeted at the industrial sector.
The artificial intelligence algorithm you developed has certainfeatures—what are they?
There are three. The first is that our algorithm has the ability to adapt to the dynamics of the context in which it is used—in other words, to generate AI on the spot and in real time. Being able to learn in a matter of seconds allows it to adapt to changes, particularly unexpected variations such as background noise in the case of a voice command, and thus continue to identify the person speaking and giving a command. The second unique feature of our algorithm is that it learns from very little data, and a single interaction with a person is enough to correct the AI if it makes a mistake. The AI will take this information into account, whether positive or negative, and reinforce its knowledge. Its third unique feature is that our AI does not require a server, the cloud, or infinite resources to perform its calculations. It can be embedded in the device and is therefore autonomous. This means it can be installed in your car, on a calculator, or on a small electronic card…
What markets is your technology aimed at?
First, there is the Industry 4.0 market, which addresses quality issues such as anomaly detection and predictive maintenance. We are currently working with Renault on these issues. We analyze the sounds produced by machine tools during operation to identify potential anomalies, and we are currently expanding our solution to include image analysis. The second sector we are targeting involvesadvanced human-machine interfaces, which replace the keyboard with gestures to control devices such as screens.
What products do you plan to sell ?
We are introducing DavinSy, an embedded software system, and Maestro, the tool used to configure it. Our goal is to sell licenses and thereby empower our customers—manufacturers of industrial IoT devices—to develop AI independently.
What challenges does Bondzai face today?
The first challenge is financial: we are currently preparing our first funding round, targeting €3 million. To do this, we need to convince the financial community that our technology—which breaks away from the dominant AI model—is a powerful driver of progress. Once the funding round is complete, we aim to expand our team to nineteen employees. Our second challenge is to ensure the product is market-ready by late 2023 or early 2024, so we can quickly generate recurring revenue. If all goes well, this diversity will enable us to expand internationally and into new industrial sectors as early as 2026. Our software solution is a fundamental, disruptive innovation at the heart of many devices, which means our market is global.
The issue of AI has sparked a great deal of concern, particularly since the wave of excitement triggered by ChatGPT. How do you respond to this?
The topic of AI stirs up fears, leading to a complete lack of understanding. Our AI is designed to produce higher-quality products while valuing human knowledge and expertise. As part of our quality control process, our AI interacts with operators, who can correct and improve it by letting it know when it makes a mistake. Our goal is to put people at the center of the learning process so that they can pass on their knowledge—not just to the machine.