WEBVTT

1
00:00:00.000 --> 00:00:01.860 line:1
BFM Business Partners

2
00:00:01.860 --> 00:00:08.120 line:1
BFM Business presents Paroles d'entreprises, with Vincent Touraine.

3
00:00:08.120 --> 00:00:13.420 line:1
A Swiss start-up based at EPFL, the Swiss Federal Institute of Technology in Lausanne,

4
00:00:13.420 --> 00:00:17.120 line:1
that helps companies digitalize their products and services

5
00:00:17.120 --> 00:00:22.920 line:1
by integrating AI tools that let them speed up their processes.

6
00:00:23.100 --> 00:00:28.180 line:1
That start-up is Omniscient, and our guest Nicholas Brandstätter is here to tell us about it.

7
00:00:28.180 --> 00:00:29.500 line:1
Hello Nicholas. Hello Vincent.

8
00:00:29.660 --> 00:00:32.940 line:1
You're a co-founder of Omniscient. Thank you for joining us on Paroles d'entreprises.

9
00:00:33.460 --> 00:00:37.760 line:1
So, Omniscient helps companies integrate AI. How do you go about it?

10
00:00:37.760 --> 00:00:41.220 line:1
Well, we noticed that the majority of artificial intelligence projects fail,

11
00:00:41.220 --> 00:00:45.860 line:1
but it's not because of the AI. It's because the digital systems that support the AI

12
00:00:45.860 --> 00:00:49.940 line:1
aren't robust and solid enough; they're often overly complex.

13
00:00:50.660 --> 00:00:54.440 line:1
What Omniscient does is build solid foundations, simple solutions

14
00:00:54.440 --> 00:00:59.320 line:1
for automation and artificial intelligence integration in an optimal way.

15
00:00:59.320 --> 00:01:02.240 line:1
I see. What sets your approach apart?

16
00:01:02.240 --> 00:01:05.800 line:1
What sets us apart is our expertise working with physical businesses,

17
00:01:05.800 --> 00:01:09.100 line:1
which are often tied to on-the-ground constraints.

18
00:01:10.180 --> 00:01:13.160 line:1
That is, instead of hiring us to carry out the project on our own,

19
00:01:13.160 --> 00:01:16.840 line:1
without looking at the tools they use, their staff and their problems,

20
00:01:18.040 --> 00:01:21.340 line:1
we take an interest, we remove that friction, we come on site,

21
00:01:21.780 --> 00:01:23.100 line:1
we simplify the tools already in place,

22
00:01:23.840 --> 00:01:26.080 line:1
we add trackers to collect the data,

23
00:01:26.760 --> 00:01:28.620 line:1
because data is very important for AI.

24
00:01:28.760 --> 00:01:30.700 line:1
Without data it doesn't work, it can't work.

25
00:01:31.460 --> 00:01:33.860 line:1
In our case it works directly with the company's data,

26
00:01:34.260 --> 00:01:34.980 line:1
the historical data,

27
00:01:35.720 --> 00:01:37.980 line:1
and that allows for a smooth, efficient integration,

28
00:01:38.240 --> 00:01:39.480 line:1
and then adding other tools afterwards.

29
00:01:39.820 --> 00:01:44.220 line:1
I see. Can you give us an example of a project you've worked on recently?

30
00:01:44.220 --> 00:01:46.420 line:1
Yes, absolutely. Recently we worked with Lasered,

31
00:01:46.640 --> 00:01:47.840 line:1
and we continue to work with them.

32
00:01:48.300 --> 00:01:52.540 line:1
It's a fairly innovative company that blends laser tag and escape games.

33
00:01:52.740 --> 00:01:53.040 line:1
I see.

34
00:01:54.260 --> 00:01:57.700 line:1
They have several franchises in Geneva and the canton of Vaud,

35
00:01:57.900 --> 00:01:59.240 line:1
and want to keep expanding.

36
00:01:59.280 --> 00:02:01.340 line:1
Yes, these are physical places where you go to play laser tag.

37
00:02:01.360 --> 00:02:06.420 line:1
Exactly. They contacted us to integrate an artificial intelligence solution

38
00:02:06.420 --> 00:02:10.900 line:1
to grow their revenue by optimizing the number of appointments,

39
00:02:11.140 --> 00:02:11.720 line:1
of bookings.

40
00:02:11.720 --> 00:02:17.100 line:1
Yes, because before, it wasn't very clear on their website.

41
00:02:17.400 --> 00:02:20.120 line:1
That's right, it was too complicated; you didn't know how to go about it.

42
00:02:20.340 --> 00:02:21.840 line:1
What we did was simplify the tools,

43
00:02:22.020 --> 00:02:26.640 line:1
we added trackers so staff can track customer data themselves.

44
00:02:26.860 --> 00:02:27.160 line:1
I see.

45
00:02:28.000 --> 00:02:31.260 line:1
The important data to track, for example, for a booking,

46
00:02:31.560 --> 00:02:35.980 line:1
is the booking they made, the booking history, the time, the place,

47
00:02:36.580 --> 00:02:40.480 line:1
but not only that. Once the customer is on site, did the person spend money?

48
00:02:40.480 --> 00:02:41.800 line:1
The revenue tied to their spending.

49
00:02:42.000 --> 00:02:44.540 line:1
All of this data matters, and we store it in a database.

50
00:02:45.280 --> 00:02:49.980 line:1
We built an AI connector that reads the data and executes actions.

51
00:02:50.340 --> 00:02:56.400 line:1
I see, so beyond simplifying the interface, there's also the use of the data behind it.

52
00:02:57.360 --> 00:03:00.000 line:1
How exactly do you guarantee the security of your clients' data?

53
00:03:00.000 --> 00:03:02.520 line:1
Data security is a very important point at Omniscient.

54
00:03:03.480 --> 00:03:07.740 line:1
First, we look at what kind of data it is, whether it's sensitive data.

55
00:03:08.080 --> 00:03:13.160 line:1
If it's sensitive data, we use sovereign internal infrastructure.

56
00:03:13.360 --> 00:03:19.400 line:1
So local hosting, a local server; we run the artificial intelligence model locally

57
00:03:19.400 --> 00:03:23.380 line:1
to prevent any leak of data from the company and the country it's located in.

58
00:03:23.700 --> 00:03:23.880 line:1
I see.

59
00:03:24.220 --> 00:03:27.500 line:1
Another important point, as you know, AI can carry out unauthorized actions.

60
00:03:28.820 --> 00:03:30.700 line:1
It can use data in the wrong way.

61
00:03:31.580 --> 00:03:33.480 line:1
To address this problem, we build connectors.

62
00:03:33.700 --> 00:03:38.240 line:1
We force the AI to go through the connector to read data and execute actions.

63
00:03:38.480 --> 00:03:38.780 line:1
I see.

64
00:03:39.080 --> 00:03:42.660 line:1
So that data is used in a secure way.

65
00:03:43.000 --> 00:03:46.420 line:1
Good. And finally, what's your ambition for Omniscient in the years to come?

66
00:03:46.420 --> 00:03:50.660 line:1
Well, today, most artificial intelligence services are behind screens.

67
00:03:50.820 --> 00:03:53.540 line:1
That is, text chats like ChatGPT, Claude or Grok.

68
00:03:53.760 --> 00:03:53.920 line:1
Yes.

69
00:03:54.640 --> 00:03:59.860 line:1
Omniscient's conviction is that, little by little, AI will move into physical systems.

70
00:04:00.060 --> 00:04:02.620 line:1
There are embedded systems, in production tools, machines.

71
00:04:03.440 --> 00:04:08.280 line:1
Our ambition is to become a leading player when AI leaves screens to settle into the real world.

72
00:04:09.000 --> 00:04:10.100 line:1
Because that will happen one day.

73
00:04:10.260 --> 00:04:13.920 line:1
Thank you very much, Nicholas Brandstätter, for being with us on Paroles d'entreprises.

74
00:04:14.000 --> 00:04:15.920 line:1
You're the co-founder of Omniscient. Goodbye.

75
00:04:16.240 --> 00:04:17.100 line:1
Thank you. Goodbye.