- il y a 1 semaine
A Conversation with Marc Benioff
Catégorie
🤖
TechnologieTranscription
00:00Sous-titrage Société Radio-Canada
00:09Are you well?
00:11Everything is fine?
00:13Yeah.
00:14Please, a very warm welcome to my friend.
00:18Cheers.
00:19Cheers.
00:19Cheers.
00:21Marc Benioff.
00:26It's incredible to have you in Paris.
00:29Ah.
00:30Yes.
00:30Thank you.
00:31Please, have a seat.
00:31Okay.
00:32I don't know if everyone knows you, but for the people who don't,
00:39and maybe there is one or two persons in the room,
00:43I have to say that you are not only a visionary entrepreneur,
00:48because what you have done is quite incredible, but as a co-founder
00:54and CEO of Salesforce, you made it the world's leading CRM company.
01:02Oh, wow.
01:02Thank you.
01:04But beyond that.
01:05Thank you very much.
01:06And under the leadership of Marc, Salesforce has grown from a small
01:12startup to a global powerhouse.
01:15We will ask you about that a little bit.
01:19Thank you.
01:20And Marc, impact goes well beyond just building an incredible and
01:26successful company.
01:27He is a champion of the idea that businesses can be a force for
01:33positive change.
01:34Yes.
01:35Yes.
01:36Yes.
01:37Yes.
01:38Yes.
01:38Yes.
01:39Yes.
01:39Yes again.
01:48Yes.
01:50Yes again.
01:51Yes.
01:52Yes.
01:52Yes.
01:52Mark is a strong advocate for equality, social justice,
01:56Yes.
01:57Yes.
01:57Promoting equal pay.
01:59Yes.
02:00Yes.
02:00Mesdames, equal pay, AGTBQ rights, and gender equality in the workplace.
02:07Yes.
02:08Marc leadership style is characterized by two things.
02:12He thinks big, and he executes right and fast.
02:17Marc, I'm honestly, truly or not, to have you here.
02:22I know that it has been difficult to get you.
02:26You are Chevalier La Légion d'honneur, absolument.
02:30Oui, oui.
02:33Vive la France.
02:39I would like to start with one question.
02:43When I went to see you some 16 or 17 years ago in San Francisco,
02:50it was a nice operation.
02:52Not bad.
02:55More than a startup.
02:57Thank you. You're very generous.
02:59And now, tell us what it is.
03:03Well, first of all, how's everybody doing? All right?
03:06All right. Okay.
03:09Good.
03:10I want to thank Maurice for inviting me here.
03:13And we've known each other a long time, almost two decades.
03:16Almost.
03:17And, of course, I love France.
03:18And I'm very honored to be now the Chevalier, the French Legion of Honor.
03:23It was a very nice surprise from President Macron earlier this year.
03:27Where I live in the South Pacific is very close to a French territory, which is Tahiti.
03:32And we did a lot of work there during the pandemic.
03:35And I was very happy to be able to support them.
03:39It was very much an honor for me to work with them.
03:42And now, it's very happy to be in Paris.
03:46And it's really the first time that I've been here since the pandemic.
03:48And I'm very happy to be back.
03:51Now, you promised that you will be back.
03:53And you are delivering against your promises, which is one of the strengths of your personality.
04:00Thank you.
04:01Well, we see each other quite a bit.
04:02Yes.
04:03All over the world.
04:03And I would say that it's been an exciting journey.
04:07And I want to thank you, first of all, Maurice, because you have been a great friend of mine.
04:11You have done amazing things with me.
04:13Also, with Salesforce, you have been a great friend to Salesforce.
04:17You bought, of course, this great company, Sapient, while you were the CEO of Publicis,
04:23which became a tremendous partner of Salesforce.
04:25So, I want to thank you for that as well.
04:28Thank you.
04:28But, after all the thanks, can you tell us what Salesforce is today?
04:36Salesforce today is the third largest software company in the world.
04:40It will do about $35 billion in revenue.
04:44And it is about 70,000 employees.
04:47We're the number one CRM, as you know.
04:50And we're one of the largest software companies here in France.
04:53And we have a lot of great employees here in France and customers here in France.
04:59All right.
05:00And we also, not just that, but also we do a lot of philanthropy and charitable giving here in France
05:07is very important.
05:08You probably know we opened the refratorio here in Paris.
05:12You came here, yes.
05:14I have been with you.
05:15We had a great dinner there.
05:16And, of course, we did that with Massimo Batturo of the great restaurateur and JR, the great photographer, who's a
05:23good friend of ours as well.
05:25And it's been an honor to give back to France, not only to be a software company here as well,
05:29and to Tahiti, the great, the best part of, the best part of France.
05:35Yeah, but I know that, Lynn, your wife loves to spend some vacation in the south of France also.
05:41Yes.
05:42So you are a great friend of...
05:43Of course.
05:44Of course.
05:44My whole life, I love the south of France.
05:47Who doesn't love the south of France?
05:49I don't know.
05:50No.
05:50I have not yet met that person.
05:52Right, of course.
05:53It's the best.
05:54One day, maybe.
05:56Moving back to Salesforce, you have said that we are going through an AI revolution.
06:03Yes.
06:03Never heard about AI.
06:05But, okay, if you say that, it's probably that it is true.
06:09Well, I know.
06:10I think I taught you how to spell AI, Bernard.
06:12Yes.
06:12Isn't that right?
06:14So I would say number one is that...
06:17You like that.
06:18Okay.
06:19Yeah, you like that.
06:19I love it.
06:20I know.
06:20You do.
06:21You do.
06:22I got one laugh out of you.
06:23It has to be also fun.
06:24One laugh out of you.
06:25It's hard.
06:26So here's the thing.
06:27Number one is that about 10 years ago, I really went out...
06:33I kind of had an existential freak out that AI revolution was going to happen.
06:38And with Salesforce, we bought a number of companies.
06:42We hired a lot of amazing AI engineers.
06:45There's a tremendous amount of artificial intelligence talent in the Bay Area where we're headquarters
06:49in San Francisco, especially at Stanford University.
06:53and we built the first version of artificial intelligence for CRM.
06:58Today, we are the number one AI CRM.
07:01We do one trillion transactions a week on our Einstein platform.
07:06You know Einstein?
07:07I heard about Einstein.
07:10Einstein.
07:11I know.
07:11Here in Europe, I have to call it Einstein.
07:13Yes, but you can say Einstein or...
07:16Einstein.
07:16Einstein.
07:16Do you understand that you have created Einstein GPT?
07:19Ah, thank you for saying that.
07:21So number one is these trillion transactions that we do every week on Einstein for our hundreds
07:27of thousands of customers all over the world, that is very much predictive transactions.
07:34It's really about machine intelligence, machine learning, and deep learning, which were really
07:41the trinity of artificial intelligence up into the last couple of years.
07:46And Einstein has really led that way for us and really in enterprises.
07:51And then the most interesting thing that's happened is as deep learning was able to have a much
07:58larger platform, it could expand.
08:00The neural networks became so much more expansive.
08:03Really about 2018, 2019, that is when we started to see kind of these breakthroughs in now
08:12what we call generative AI, and that idea now that we have generative artificial intelligence
08:19within Einstein is really going to be the democratization of this amazing technology
08:25for all these enterprises as well, just as the previous generation was so important for them.
08:31But there's a difference in how we do it.
08:34We are about delivering trust.
08:36For our customers, the top banks here in France, the top insurance companies,
08:42the top media companies, the top technology companies, and so forth, their data is the
08:49most sacred aspect of their business.
08:52But the way AI works is AI wants the data.
08:58So that's kind of like ChatGPT.
09:00ChatGPT is like a huge vacuum cleaner that's trying to take all the data off the internet, learn
09:07all the data from any site it can get, scrape every site.
09:11And then consume that data, and then it delivers its intelligence.
09:17And I'm sure you've used the product and has a natural interface on it.
09:21To be honest, I have asked ChatGPT to do the introduction for you.
09:27Yes.
09:28And when I read that, I said, oh, shit, oh, sorry.
09:31This is really bad.
09:34It is much better than what is written by ChatGPT.
09:38And it was just a recap of the history which we can find anywhere.
09:46Well, the thing that's interesting about this is that once all the data is consumed, it can
09:52then start to do all kinds of different things.
09:55It could give you a very clear and specific answer.
09:57It could write a screenplay.
09:59It could write a song.
10:00It could do a haiku.
10:02But also, it could lie.
10:05Yes.
10:06In fact, it can be a terrible liar.
10:08It can hallucinate.
10:10It can, that's a technical term in artificial intelligence.
10:14It means that even though it has all the data, it comes up with the wrong answer.
10:18Because the way it works is it's consuming all the data, and then the algorithm is this,
10:24this, this.
10:25Therefore, this must be true.
10:27But of course, it doesn't always get it right.
10:30And we've seen those examples.
10:32In the United States, this past couple of weeks, we had an incredible example where a lawyer
10:37submitted to the courts this huge case that he had built on ChatGPT and asked ChatGPT,
10:43are all these things correct?
10:45Is this the right way?
10:46Can I submit this now to the courts?
10:48But then, of course, it was all wrong.
10:50Yes.
10:50And we have seen also burn what has been created by Google.
10:56Well, I think this idea that getting to the next point of accuracy, which for a business
11:03is different than a consumer.
11:04So you know the consumer brands very well.
11:07And when I'm having a consumer experience, and when I'm having a natural language experience,
11:12and when I'm just talking to ChatGPT like you're doing, that's one level of trust.
11:18And maybe it's not as important.
11:20But when it's the business, and I'm having an interaction with my business, and I'm talking
11:26to it, then trust is going to become paramount.
11:29And we must get to another level of trust.
11:32So the way that we have built our capability, which is, yes, we have the best generative AI.
11:38You can see we have models we've built.
11:40We publish our models on open source and so forth.
11:43But we've built a trust layer.
11:47So we never look at your data.
11:49We don't have to look at the customer's data to give you the powerful generative experience.
11:56This is going to be so important for our customers.
11:59Because just in predictive, we never looked at the customer's data.
12:03We were able to tokenize the data and keep it anonymous to us, private, so that we can then
12:09deliver these capabilities, but without looking at the corporate data.
12:14This is what our customers truly want.
12:17They do not want to amalgamate their data.
12:19Let me make that clear.
12:20The way this works is, these large language models, so really machine intelligence in general,
12:26but especially the large language model, it amalgamates the data.
12:30It wants all the data, and then it has to pack it all in, and then it can kind of
12:36come up
12:36with the answer.
12:37But that's not how the data works with our large customers.
12:41We have sharing models.
12:42Not every executive can see what every other executive knows in a company, as you know.
12:47It has to be very specific for every executive.
12:51That is why the relational database model has been so successful in business, because you
12:56have security down to the cell level, to the row level.
13:00Readers don't block writers, and I know that this cell is protected, and only I as a user
13:06have access to that information.
13:08But that's not the way that generative AI works.
13:11Generative AI works in an amalgamated way.
13:14So we have to use the amalgamated technique, but we have to anonymize the data, and then we
13:19can deliver trust back to the customer.
13:22And that is going to be the real power of what Salesforce will be able to do.
13:26And we're going to release the new generative AI versions of all of our products in the
13:32next week or two.
13:33So we just in New York on Monday had our AI day where we showed our new sales GPT, service
13:41GPT, marketing GPT, commerce GPT, on and on and on for all of our products to show that
13:46are customers who will automatically get this release and be able to start using this capability.
13:52And then we will incrementally, every month this year, start to release new versions of
13:57our generative AI.
13:58Because our customers rapidly want to move from where they are now, which is very good predictive
14:05artificial intelligence, to higher productivity.
14:08But not at the cost of trust.
14:13So this aspect will be critical, especially here in France, where privacy is paramount.
14:21Where do I sign?
14:23Because when I hear you, you are so…
14:25Well, thank you for actually mentioning…
14:26You are so…
14:26Yeah, I have it right here.
14:27Oh, okay.
14:28All right.
14:28You are so convincing that I say, okay, I have to sign immediately the contract.
14:34It's quite impressive.
14:36This is the reason why you called it Einstein?
14:39Well, no, I think that this is very important because for our customers, because we work on
14:43so many companies together, don't we?
14:44And many of them are right out here, like LVMH, BNP, right?
14:49So many of our great customers here in France, AX, okay, on and on.
14:54But their data, every employee cannot see what every other employee sees.
14:59It's very specific.
15:01That's not the way generative artificial intelligence wants to work.
15:05It wants to amalgamate.
15:06It wants to combine.
15:07It wants to aggregate.
15:09This is the power.
15:11That's why ChatGPT is out there with its vacuum cleaner, right?
15:14And then it's just putting it all into a large language model.
15:17And the funny thing is, Maurice, I just had a, I was in New York on Monday, and a bank
15:23CEO, one of the largest in the world, we're talking, oh, I'm going to take all my customer
15:27data, I'm going to put it in the large language model, I'm going to increase the productivity
15:32of my company.
15:33I'm like, yes, you want to do that, but you're not going to put all your data just blanketly
15:39into a large language model.
15:40All the account balances, all the customer names, all the transaction history, and then
15:45anyone can see it.
15:46No.
15:46Or health data.
15:47All of your health data is going to go into a large language model.
15:51No.
15:52That's not the way we're going to do it for enterprises.
15:55We have to think about enterprise and consumer differently.
15:58We all want the productivity or the intelligence or the amazing experience like you've had,
16:05we've all had, you know, with the NLP interface, you know, the text interface, but we must have
16:11it with trust.
16:12This is going to be the number one most important thing that we all must strive to find in the
16:17enterprise so that we can have this capability.
16:20You have just seen the power of conviction, which is quite unique, the power of persuasion
16:27that Mark has, and this is leading me to another question.
16:33So you moved from AI, and now you are speaking about AI cloud.
16:41Yes.
16:42So what it is, it is the complement to Einstein?
16:47Yes.
16:48Well, I think to get these capabilities, it's going to require several steps for customers.
16:55One thing is, we're going to have to build our data cloud.
16:59Every company is going to need a data cloud in their CRM customer data, like they have
17:04with a Snowflake, or a Databricks, or a Google BigQuery, or some of these types of things.
17:10Clean room.
17:10Yeah, but they want it integrated with their metadata also.
17:15So the metadata must also be tightly woven into the data, and that has to be the data cloud.
17:22That's going to require the capabilities of the generative AI to really have those insights
17:28will be the integration of the metadata and the data together.
17:31Yes.
17:32That data cloud, plus we also have a company called Tableau.
17:37Yes.
17:37It's a French word.
17:39It's an international word, I believe.
17:42Tableau.
17:43This is one of our great companies that does also the data analysis and visualization.
17:49We also have integration with MuleSoft, and we even have Slack, which is giving you this
17:55kind of interface where you need a secure text-based interface with the generative AI.
18:00I think that Slack will be probably the most successful interface with generative AI.
18:06You can already see companies like Slack, like OpenAI, companies like Anthropic, which
18:13is another large language model, vendor Inflection.
18:16All of the top AI companies are already Slack companies.
18:20Because of that capability, they're able already to take the data from Slack because it's an open
18:26data model itself to generate this next level of intelligence.
18:30So we have to deliver a complete solution to our customers, and all of these things are
18:36going to form our AI cloud, which we're now delivering to our customers, and we're very
18:40excited about this.
18:41So I understand now a little bit more about the success of Salesforce, the incredible success.
18:48You are always for moving to something which is not yet there, beyond the hill, and you see
18:56what will happen tomorrow.
18:59One of the things for which you are extremely famous is stakeholder capitalism.
19:07Yes.
19:08Maybe you can say one word about it.
19:10Sure.
19:11Because you are showing that being a good company, doing good business doesn't mean that you don't
19:22have to be also good for the other one.
19:25Yes.
19:26Okay.
19:26Well, you know, we started our company 25 years ago, and when we started our company,
19:32we put 1% of our equity, 1% of our profit, and 1% of all of our employees'
19:37time into philanthropy.
19:39Okay.
19:40It was very easy, because we have no employees.
19:43We had no equity, no capital, but today it's different.
19:48So we've given away more than 600 million in grants all over the world.
19:53We run 50,000 nonprofits and NGOs for free on our service.
19:58And we also have done almost 8 million hours of volunteerism wherever we go in the world.
20:08We're going to give back.
20:09Like here in France, we've done so many amazing projects in the local communities.
20:13I was just hearing from some of our employees about amazing things they've done.
20:16That's a very important part of our culture that we call ohana, the giving back, really
20:22trying to support others to strive for greater equality.
20:27Or like we say here in France, equality and fraternity and liberty.
20:31That ability to have values like that built into a company is so important today.
20:36When we talk about trust, like with AI, customer success that you and I have focused on so many times
20:43for our whole career.
20:44Innovation, equality and liberty and sustainability.
20:48These are the things that are important to Salesforce.
20:52So that is stakeholder capitalism.
20:55We want to make sure that we're not just about doing well, but doing well and doing good.
21:01And when we have grown our company, we've worked hard to make sure that we always do well and do
21:07good wherever we go.
21:08I think every company can do this.
21:11I think it's a lot of the fabric of France already.
21:14A lot of the companies here already understand this very well.
21:17They do it.
21:18But this idea that we're striving for not just our shareholders, that the business of business isn't just business,
21:25but the business of business is improving the state of the world.
21:29And that every business can do this.
21:32And business has to see itself as an integral fabric of society, not independent or separate from others,
21:41but that we're all connected as one.
21:45Mark, you are a master.
21:48And when I'm listening to you, I see what is driving Salesforce so brilliantly.
21:58and why you are so successful.
22:01So I would like to thank you.
22:03My mother also feels the same way, Maurice.
22:05So I appreciate, you know, you and my mother feel a very similar way about that.
22:11So I appreciate this.
22:12She told me also that you have to improve.
22:14She told me that you have to call her more often.
22:17Yes.
22:18My mother lives with me.
22:19So that's not a problem at all.
22:21My mother's 84 and she lives at my home.
22:23So I see her plenty.
22:25Yes, but you are not often at home.
22:27Yeah, I see.
22:28Well, you know how that is.
22:29Okay.
22:30No, no.
22:30I'm just kidding you.
22:31I would like to thank you very well.
22:33Maurice, thank you for your leadership of France.
22:36Thank you.
22:36Thank you for your leadership of the tech industry.
22:38Bravo.
22:39Thank you for creating Viva Tech.
22:41Thank you for encouraging me to be here.
22:43And Viva la France, Maurice.
22:47merci.
22:48Thank you, taking a round.
Commentaires