00:00[Music] My name is Emmanuel Frenehard, I'm the chief digital officer for a biopharmaceutical company called Sanofi
00:08As a company, we set ourselves a pretty incredible purpose, which is to chase the miracles of science to improve people's lives
00:21Saving time is the one thing that we can do, because time is a currency that most patients don't have
00:27So we said, how can we improve the lives of millions of people around the world, and what could we do? And technology and AI just came at the perfect meeting point
00:38If you look at our cycle, the cycle of developing a drug in our industry is very long, it can take 12 years or more to get from an Idea, a hypothesis to a treatment, 90% of drugs unfortunately fail, so 12 years with a 90% chance of failure, those are really tough numbers
01:00And so if we can use technology, and we can use artificial intelligence to change that and shave time in those 12 years and increase the probability of success, then we all benefit from [Music]
01:15Four years ago, when I joined Sanofi, it was quite a different company
01:18The company had uses of artificial intelligence, but they were sporadic and they were not coordinated as such
01:23You could see in our factories, you could see in our labs, you could see in parts of the company
01:29Company, but we decided that we wanted to bring everything together and create a company-wide strategy, truly an enterprise vision behind artificial intelligence
01:39And with that vision, we really came up with three categories of artificial intelligence that we want to differentiate ourselves
01:49The first one is expert AI, the AI that you find in our labs, in our factories, the AI that is truly designed for the expert, those that will use it
01:57If you're looking for a cell, for example, we have an AI for it
02:04If you're looking for a deviation, we have an AI for it
02:12The second category, and the term was coined by our CEO, is Snackable AI, kind of like your Instagram of AI, if you want, where you have at your fingertips recommendations that are available to you in an app that we co-created with a company, um, called A-Labs
02:19That kind of technology is now available to thousands of employees
02:25Over 22,000 people inside Sanofi have access to it, and they can make decisions at their fingertips
02:29And the third category, and I think most of us have been exploring that category these days, is generative AI
02:35There's a certain creative ability to generative AI, so we're able to create new content, localize it at speed, and reach patients
02:45Healthcare professionals faster than ever before
02:51For our employees, we've launched something that we call "concierge" at this point in time, which is kind of this idea that there is AI and elements of digital help that you can get to do activities that you really don't have to do on a daily basis
03:05So we want to be able to book your holiday, inform your team when you're out, book days, do activities like that through a concierge, and really a digital concierge, which is full of generative AI
03:19That's been our approach
03:22I think that approach will happen in all our lives, and we'll all have some sort of a conc-, or Multiple conures that will help us for our day-to-day, every time you launch something new and you bring a lot of innovation, of course, you're going to have some setbacks
03:33There's moments of failures, and those failures are the moments of corrections
03:36What we've learned is, first of all, the technology wasn't quite as ready as we had anticipated it, and so that was a, that was a humbling lesson
03:44The technology also evolved much faster than we anticipated
03:47When we had something fairly stable that we thought was going to deliver value quickly, a new technology was available that superseded it, so you've got To be wise about innovating at speed, but also be willing to sacrifice what you just developed, because something better comes just after
04:06When you're looking at implementing AI, there's a risk of doing a little bit too much and really not going deep, so we saw that we wanted a strong governance
04:15We created a gen board, and in that governance, we select use cases, or value theories as some call them, that have a tremendous amount of potential impact for the company, whether it be creating Top Line, shaving time, just the impact is an element of selection, and then we agree that we're going to go deep, we don't
04:36Want to sprinkle dust, we just want to go deep and solve a problem from the ground up, and truly deliver on the value that it can generate
04:45That's been, I think, one of our recipes of success, and that I think has been our biggest lessons learned in generative AI, just focus on what value you're trying to generate, ensure that you have the right level of technology, test test benches, and then deploy it as fast as possible in your production environment so you can learn from it
05:06I think too many people shy away from putting things into the real world, POC's, proof of concept prototypes are shallow, they
05:16Don't deliver value, they deliver hypothesis testing
05:22The hypothesis in the real world, with minimum viable products that you put in the hands of your users, your clients, your patients, is far more powerful.