The adoption of artificial intelligence in companies often requires support. In this issue 3 of Synapse AI, Mathieu Barreau meets Julian Lucchesi, Director of Strategic Partnerships at Centech, a Montreal-based startup incubator specialized in DeepTech.

The incubator offers an innovation hub for SMEs and large corporations to help them adopt AI. This interview is an opportunity to discuss critical issues such as project structuring.

Testimony of Hugo Gagnon, CEO of PhenoSwitch BioScience, a company that passed through the AI HUB of CenTech Montreal.

Hugo Gagnon, you are the General Manager of PhenoSwitch BioScience in Quebec, you have just done the AI HUB of CenTech Montreal, what do you do within your company?

PhenoSwitch BioScience is a life sciences contract research company founded in 2002, whose day-to-day business is to analyze samples for our clients. We perform Omics analysis, using mass spectrometry coupled with liquid chromatography. We can separate molecules, identify them, sequence them and quantify them. Our activities allow us to follow the mechanisms of action and to better understand the efficacy of molecules and we generate a lot of data.

Our challenge is to analyze this data better, more quickly, more completely and more reproducibly. It is in this context that we have turned to machine learning.

Was Artificial Intelligence already present in your activities before you created the AI Hub?

Not really, no, at least not artificial intelligence as we use it today. We were already using some scripts for data analysis, some methods (PCA, clustering etc…). But we needed to go further, to be able to predict classifications, to predict the behaviour of molecules, and that’s what we can do now thanks to Deep Learning, and in our field it’s just starting to be used.

When you access the information available today, you can find publications or tools available but there are often B-mol. We really wanted to find solutions to the b-mol that we had identified in different scripts.

How did your adventure at the AI Hub begin?

I was able to attend a presentation of their method and approach, I was quickly seduced because we have developed the same relationship with our customers, when we evolve in contractual and complex areas, there are often two scenarios, either the customer knows what he needs from the beginning without knowing the risks or difficulties related, or the customer feels that the solution exists and that it could meet his needs, but he does not know how. And this is often the case when we talk about using artificial intelligence. We know that’s where we need to go, but we don’t always know the best way to get there. The AI Hub allowed us to ask ourselves questions we hadn’t thought of. We mapped our activity with a BMC, which was a way of making a shared assessment, asking ourselves the right questions, thinking about and defining our solutions and prioritising them. Then we did a proof of concept, which showed that it was possible to do what we wanted to do and that it worked.

What was the reason or reasons for wanting to use AI within PhenoSwitch BioScience?

We developed the answer to our initial problem, which can be found in the scripts currently proposed, they lack precision, they are often too binary and theoretical and do not take enough into account the exceptions. A concrete example, we have to analyze proteins, they have to be digested, we use enzymes for that and it is in this precise case that the theory does not allow any exception, whereas it does not reflect the reality. Until yesterday, the intermediate solution was to generate data, produce our ionics libraries and integrate the data. We are talking about 3 to 4 weeks of work, which is not insignificant for an organization.

Have you encountered any unexpected things during your journey with the AI Hub?

We underestimated the time at the beginning, because we had to explain our job, what biology is to AI experts, we had to learn to dialogue on the same language.

I think that when we arrived at the AI Hub with an already identified problem, but in the end we came out of the program with more than the answer to our problem, we have a more macro perception of the possibilities that are available to us. We became aware of the potential of AI applications in our daily lives, of all the tasks and operations that could be automated thanks to AI.

It also changed our IT structure, we had to centralize information in order to be able to exploit it properly, this represents investment costs, but they are necessary.

Today, we have to invest in staff training so that everyone can use the tools we have developed. It also seems unavoidable to recruit a person specialized in AI who will be able to understand the whole operation and who will be able to intervene on maintenance and train the algorithms on data sets.

Do you consider that you have an advantage over your competitors today?

Yes, it’s undeniable, we are in a business sector where artificial intelligence will be present either from the start, with very innovative startups, or very little in companies that have been developed for several years and already have processes and operations in place and do not perceive AI as an opportunity. PhenoSwitch BioScience is halfway between these two situations, we do not have the ambition to revolutionize our field, but to accelerate, facilitate and improve our organization, yes, we are more efficient. Let’s not forget that in Quebec, we are facing a shortage of skilled labour and this transformation is also a strategy of anticipation. Instead of having to clean and process data, our scientists can now focus on their real job, answering scientific questions.

Do you feel that you have been able to find sources of funding easily? Yes, certainly because as researchers we are used to putting together funding applications but it is still complex. For example, for one of our projects, the conditions involved finding another partner to join our project and it was not easy to find another company with similar or complementary needs, which would agree to share part of the IP (intellectual property) and which could finance its part. The financing mechanisms are not easy to set up but they are there.

You have achieved a first step of transformation, will you continue to integrate AI in your organization?

Yes, we will. We have two other major AI projects that will allow us to complete our added value on the market.

Translated from Synapse AI : Entretien avec Julian Lucchesi, directeur des partenariats du Centech à Montréal