Latest / Energy Voice – Out Loud / Gastech Bludigit
Transcript
- 0:05Hi, I'm Mark Selby, Editor of Energy Voice.
- 0:08I'm here backstage on the show floor at Gas Tech with Serena,
- 0:12Delhi Data and AI Architect at Blue Digit.
- 0:16Really great to be with you, Serena.
- 0:18Glad. To be here.
- 0:19So yeah, we're we're a gas tech. I think you probably just
- 0:21dropped by to get a feel for things and soak up the
- 0:25atmosphere. Like what do you think of the
- 0:27event so far? It's huge.
- 0:29I was not expecting to be this big.
- 0:33And it's a lot of people out of vendors, a lot of ideas.
- 0:38That's great. Have you heard anything exciting
- 0:40at all? And I mean, the, the, the, the
- 0:43overarching theme for this this whole event, all four stadiums
- 0:47worth is, is AI. And you know, AI is, is
- 0:50featuring almost every keynote speech and all the major
- 0:54presentations are touching on that theme.
- 0:57Have you heard anything at all that's, you know, really
- 0:59resonated with you or? Yeah, I think nowadays AI is the
- 1:05main things everywhere, so it's kind of cool for my my
- 1:09perspective, my job. I think it's very interesting to
- 1:13see how all the companies are trying to lever into AI in their
- 1:20work workload and how they are doing things differently and how
- 1:27energy is quite, it's a quite demanding thing for AI, so.
- 1:34Yeah, of course the the, the huge increases in demand for AI
- 1:38are really featuring big in in the keynote speeches from the
- 1:42likes of Shell. And you know, you can't escape
- 1:46that that trend here. It also feel, it's interesting
- 1:49how you kind of feel like these big companies are hedging their
- 1:52futures on to a certain extent on that power demand growth.
- 1:57But maybe before we kind of get into that, just tell us a bit
- 2:01more about Blue Digit and your role there and, and what you,
- 2:04you know, the journey you've been on.
- 2:05So Blue Digit is the tech company of Italgas.
- 2:09Italgas is the 1st guest distributor in Europe.
- 2:13We operate not only in Italy but also in Greece and in Italgas
- 2:20group we have different companies and I'm part of the
- 2:24Digit. The Digit offer tech support,
- 2:29tech innovation and advisory not only for the company in the
- 2:36group but also to other other player, mostly in the energy
- 2:42market. And we also develop proprietary
- 2:46solution that we sell to other to other company.
- 2:52What I do I'm the architect. Basically, I play with the AI
- 2:57and I try to leverage data and information to make the workload
- 3:05of my colleagues easier and more efficient.
- 3:08So what what is that? I mean, all the big and
- 3:13successful tech start-ups in the world have always kind of come
- 3:17from a place of having that USP. You know, there's that.
- 3:20That's the the special source. Like what is the problem that
- 3:24your company set out to solve? So, well, we mostly work with
- 3:31etal gas of course. So we know really well the
- 3:35problem of gas distribution and the worksite.
- 3:40So what we really try to achieve is making the worksite easier to
- 3:46manage and more safe. And how we are doing that, it's
- 3:51with our tool that it's called again not fantasy in there, but
- 3:56it's called worksite work on site.
- 3:58And we are, we are integrating AI to make, for example,
- 4:09approval of the work done in the work site easier.
- 4:13So you just take a picture and the AI will check if everything
- 4:18is there, everything you were supposed to do is done if I
- 4:22don't know, the pipe was placed at the right depth and then with
- 4:28the just a click, everything is done automatically.
- 4:31So you have no bureaucracy at all.
- 4:34What we are working next is the safety on the work site because
- 4:39it's quite an important task for us.
- 4:43So what we are trying to do is to collect image from the work
- 4:47site and check if every safety procedure are in place so that
- 4:52we can guarantee a better and safer work environment.
- 4:58Right. So it's, it's primarily about
- 4:59safety then. So along the way though, in
- 5:03terms of, you know, refining and improving your offering and, and
- 5:06how AI fits into all of that, like have you encountered any
- 5:11specific challenges or obstacles that you've had to overcome to
- 5:15get to to where you are now? Yes.
- 5:16I think especially at the beginning, especially when
- 5:21generative AI came into place, people were really, really
- 5:28scared about losing their jobs. So we really struggle to make
- 5:34our users feel safe enough to use the technology, but not safe
- 5:42that the technology was correct, but safe that this wasn't going
- 5:47to impact their jobs. We, we really believe that AI,
- 5:52it's not replacing humans, but they are just algorithms are
- 5:58just trying to make your workload more manageable and
- 6:05like giving you space and time to achieve whatever you want to
- 6:12do in a best way, best way possible.
- 6:16So it's not really about replacing, but augmenting what
- 6:20we can do. And yes, the change management
- 6:24was really, really tricky. And what we did was to combine
- 6:30the IT department with HR and the working together to change
- 6:38and educate people so that they could really see our vision and
- 6:44feel more comfortable in using our products.
- 6:49So it goes back to that thing that has been said a lot over
- 6:52the last couple of years about AI, which is you're not going to
- 6:56lose your job to AI, you're going to lose your job to
- 6:58someone who's using AI. Yes, I think it's really
- 7:03important especially for younger people because they ones have
- 7:08experience like work experience, but they will be able to bridge
- 7:16the gap that they have with the tool that they can use.
- 7:21Because I think a lot of people don't really use AI in the right
- 7:26way. They don't really know how to
- 7:28use it. And so it can be very important
- 7:33for your career and but even just for for your life.
- 7:38I think I'm using AI for everything nowadays.
- 7:41So it's really something that it's your gain, not just the
- 7:47company. So would you agree personally
- 7:50then with, as I was saying earlier, you know this big theme
- 7:53that's being talked up around gas tech of the huge potential
- 7:58and actual increase in power demand that's coming from AI?
- 8:03Yeah, I think we have to be wise because what I've seen is that
- 8:10everyone wants to have AI in their product, but it's not
- 8:16necessarily the best way to automate the process or to make
- 8:22it smarter or easier. It's not because it's fashional
- 8:27fashionable that we have to use it.
- 8:30I think we have to really think about when it's more convenient
- 8:34and when it will give us the best in outcomes because it's
- 8:43very high demanding in energy. So sometimes old techniques like
- 8:50machine learning or deep learning or even just automating
- 8:54a process, it works better than generative AI, which is the
- 8:59really high demanding in energy. Yeah.
- 9:04And we just don't, we have to think about it.
- 9:08We have to be responsible. So I think it's a huge concern
- 9:13that we should have and I think we are not really having it.
- 9:18That's interesting. So from your perspective then it
- 9:21there's, there's an element of caution and risk management
- 9:25rather than, I mean, obviously your company sits in a different
- 9:28space to some of the huge energy producers, you know that you'll
- 9:32find talking talking shopper at gas tech.
- 9:35But yeah, so there's a note of caution there that maybe others
- 9:39would be, would be lost kind of in the rush of enthusiasm.
- 9:43Yes, I think well, for sure we will have better energy, better
- 9:51more efficient LLMS algorithms, They will get better.
- 9:57We saw with Deep Sick that it was already way better than what
- 10:02we have. But I think we should all also
- 10:08keep in mind that it's not that we have to.
- 10:11It's not mandatory to use generative AI.
- 10:14We can use other things and achieve the same results.
- 10:19That's that's interesting coming from an AI and data
- 10:22architecture. So it's nice, a nice honest note
- 10:24of caution. But yeah, otherwise like in
- 10:28terms of what your company and and yourself personally, you
- 10:31know what you're looking to towards in the future, where do
- 10:35you see like the next milestones or opportunities coming?
- 10:38I think it will be really, really interesting when small,
- 10:43small language models will will be really usable, because
- 10:52nowadays they're more experimental.
- 10:54But I think they will make a very big change in generative AI
- 11:01and even in the energy demand because they will be low,
- 11:06smaller, so it will be easier to train them and to use them.
- 11:12So I think that will be a very big change in the field.
- 11:16And for your, for your company specifically and for for your
- 11:20service is that, I mean you're talking more generally, but is
- 11:23that something that will apply to blue digit?
- 11:25Yes, of course, of course it will.
- 11:28It will apply to all our prod, our products, because what it's
- 11:32the problem now is that all the products are online.
- 11:36So you have to have signals, some kind of signal.
- 11:41But when you will be able to use a small language model, you will
- 11:47have all the algorithms running in your device and it could be a
- 11:54telephone to a mobile phone, so a very small device.
- 11:59It doesn't have to be a computer with a big memories and
- 12:04processors. So it will be really interesting
- 12:08to see how you can expand your products and how many things you
- 12:15will be able to achieve. So if you were looking ahead to
- 12:20say five years time, not to be too cliched, but five years time
- 12:24or 10 years time, like what would you like to say that you
- 12:26and Blue Digit had achieved? So for sure having all the
- 12:31safety tools and in place, this is something that we will, we
- 12:37hope we will have in the very short future.
- 12:44So what we are trying to do as well is to build a more reliable
- 12:49ESG platform where we can not only control how much energy we
- 12:55are using in our buildings, in any buildings like the waste
- 13:03productions or this kind of stuff, but also the CO2 emission
- 13:11that we have with our products and with anything related to the
- 13:18cloud Internet providers. So with emails and calls and
- 13:26documents everything will will be traceable and we will know
- 13:31how much is going to cost environmentally talking.
- 13:37And finally, I'm just really interested to get your opinion
- 13:40on this. Do you think that the future of
- 13:42decarbonisation and reaching net zero is reliant on AI?
- 13:47Do you think that that is the way we're going to achieve it
- 13:50globally? I think AI is going to help.
- 13:54I think AI is going to help in every single field.
- 13:58It's not. It's not something field
- 14:03specific, it's something that can work with everything, but I
- 14:09don't think AI is enough. So we really need to put the
- 14:14effort and the work and the willingness to reach net 0.
- 14:19Otherwise we won't. It's not AI that is that is
- 14:22going to save us. Some of the suite of products
- 14:25and services that you and Blue Digit offer, like do you have a
- 14:29favorite? Like, is there anything that
- 14:31kind of jumps out at you as something that's a particular
- 14:33particularly exciting or interesting?
- 14:35So it's a very it's a quite simple algorithm, machine
- 14:40learning, so not generative AI know something super new.
- 14:44But I think it's really interesting because I didn't
- 14:49know that gas doesn't have the smell before join Italgas.
- 14:56But basically what you smell the the the smell of gas is actually
- 15:05something that is created is a chemical substance and
- 15:12authorizing substance that is put in the gas for safety
- 15:18reason. And we have an algorithm that
- 15:22check if the level is is if we have if we reach the safety
- 15:30level for the gas in all the steps in the pipelines.
- 15:37And this is something is important because it's maintains
- 15:43a level of authorization that allow human to smell a leak.
- 15:48So it's really important for safety reason and also what the
- 15:53algorithm does is to optimize the recharge of the other
- 15:59authorizing product. And it also does something
- 16:05important for the economic part of for the company, which is its
- 16:11schedule when to buy the new authorization products according
- 16:18to the price drops. So it's really interesting and I
- 16:23really didn't know that gas doesn't have a smell.