Agentic AI - Use Cases, Benefits, Risks - Webinar Preview

WEBVTT
00:00:06.889 --> 00:00:10.955
All right, welcome everybody to the SNEA Experts on Data podcast.
00:00:10.955 --> 00:00:15.131
This is amazing because we've got another fantastic webinar that's coming up.
00:00:15.131 --> 00:00:29.121
So this is a webinar preview and I'm super excited because I'm joined by two of my favorite humans who are going to talk today about agentic AI and a really, really cool webinar which is coming up.
00:00:29.121 --> 00:00:31.568
So, first of all, I'll start with introductions.
00:00:31.568 --> 00:00:38.872
So, if you don't mind, give a quick intro where you're at what you do, and then we'll jump into talking about the webinar.
00:00:38.872 --> 00:00:39.841
We'll start with you, erin.
00:00:40.902 --> 00:00:41.061
Hi.
00:00:41.061 --> 00:00:54.337
Yes, I'm Erin Farr and I work for IBM in the storage CTO office and that's a strategy role where we look at industry trends, disruptive technologies like agentic, ai and customer pain points to influence the roadmap of our solutions.
00:00:56.100 --> 00:01:43.862
Excellent and Michael, no-transcript, because it seems like it's so new and yet it's been on the tips of tongues of CIOs and pundits everywhere for the last little while.
00:01:43.862 --> 00:01:51.075
But what I'd love to sort of start by framing, Erin, what is your goal here?
00:01:51.075 --> 00:01:51.902
I know we've got this.
00:01:51.902 --> 00:01:54.349
The webinar is really the core of what we're going to talk about.
00:01:54.349 --> 00:02:00.111
But what's exciting you about agentic AI and what's your goal with the webinar that's coming up?
00:02:01.174 --> 00:02:01.414
Sure.
00:02:01.414 --> 00:02:11.768
So my goal is to give enterprise storage personas the information they need if they're evaluating agentic AI solutions.
00:02:11.768 --> 00:02:26.201
Right, because there's a lot of agentic information out there and it predominantly targets developers people who are building agentic AI solutions, but not necessarily those who are consuming them and what to look for, but not necessarily those who are consuming them and what to look for.
00:02:26.222 --> 00:02:32.911
Well, this is always the interesting thing too, and we mentioned, you know, we've chatted before talking about the idea of like, who's the target for this?
00:02:32.911 --> 00:02:39.782
And it's interesting because when we say AI, it's kind of an everyone's game.
00:02:39.782 --> 00:02:43.051
Then we say agentic AI and it tends to lean towards sort of builders and developers.
00:02:43.051 --> 00:02:57.921
But what's funny to me you know funny, peculiar, not funny, ha ha is that we really forget about compute, storage, networking, all of the real sort of pain points that you're about to discover.
00:02:57.921 --> 00:03:05.948
If you think that just agentic AI is the be all, end all and it's fantastic and what it can do, Let me start with that.
00:03:06.639 --> 00:03:15.725
But there's big implications that are further down the stack and even higher up in the stack as far as what are the goals and business outcomes?
00:03:15.725 --> 00:03:21.947
What are the expected sort of ramp times when people are taking on an agentic AI project?
00:03:21.947 --> 00:03:25.282
What is the reality of the maturity of an agentic AI?
00:03:25.282 --> 00:03:27.118
So, what is the reality of the maturity of agentic AI?
00:03:27.118 --> 00:03:33.359
So there's so much to cover, which is why this is a preview and people will make sure that there's going to be a link below.
00:03:33.359 --> 00:03:35.066
If you haven't already signed up for the webinar?
00:03:35.066 --> 00:03:39.366
Please do, and we'll share exactly sort of details of that at the end.
00:03:39.366 --> 00:03:46.680
But from your side, Michael, on the SNEA world, how does agentic AI play into the work that you've been doing?
00:03:50.786 --> 00:03:51.325
Well, great.
00:03:51.325 --> 00:04:13.507
So we have had a number of webinars this is part of a series and we've had webinars on AI ops and went into a lot of detail about how folks prepare and then deploy into IT and different systems for enterprise.
00:04:13.507 --> 00:04:19.601
We've also had SEF, the Linux of storage, and that was very fascinating.
00:04:19.601 --> 00:04:27.473
A lot of that is being driven by the scale need for AI and that was addressed in that talk.
00:04:27.473 --> 00:04:46.423
We've also done Kubernetes a series on that for the fast moving storage and scale needs that they have, needs that they have and especially for managing all of that data.
00:04:46.423 --> 00:04:57.899
So, in terms of the Agenic AI, this is a very, very new area and we're really excited to have Aaron kick this off.
00:04:57.899 --> 00:05:02.711
This will also be a series, so we're looking forward to more of these.
00:05:02.711 --> 00:05:06.291
After the end of June We'll have our next webinar.
00:05:06.524 --> 00:05:19.206
I think a perfect lead in here, Aaron AI.
00:05:19.206 --> 00:05:22.175
And why is this an amazing time to come in and learn about what's coming as a result of this new sort of wave in?
00:05:22.254 --> 00:05:27.831
the AI world, right, and so yeah, and I plan to go into deeper detail about what it is in the session as well, but at a high level.
00:05:27.831 --> 00:05:30.846
I'll talk about it in the context of generative AI.
00:05:30.846 --> 00:05:37.273
Right, with generative AI, a large language model generates content as the result of a prompt, a question.
00:05:37.273 --> 00:05:38.384
You ask it, right.
00:05:38.384 --> 00:05:52.257
But with agentic AI, the language model now generates actions to execute on your behalf or skills, they're sometimes called, but it does these actions using tools or skills it has at its disposal.
00:05:52.257 --> 00:05:59.151
So agentic comes from the fact that it has agency to act on your behalf and even run autonomously.
00:05:59.151 --> 00:06:05.011
And I'll cover a little bit the evolution of how we got there and what the differences are between agentic AI and generative AI.
00:06:06.874 --> 00:06:18.966
And I'd like to think that this is probably almost a Gutenberg level of innovation in what we've been able to do with AI, particularly around.
00:06:18.966 --> 00:06:52.975
Like you said, this is a focus specifically on generative AI, because it was very, you know, having human in the loop was one thing that's typical in AI consumption today, but then to go one step further and to be able to, like, give it long running tasks and complex tasks and then have it actually figure out based on roles and capabilities and now we see things like the introduction of MCP and more things around how do you present and publish what services you have available?
00:06:52.975 --> 00:07:17.091
This is all happening so fast, but it really is working towards what I believe is a common, shared goal of the AI industry, in that how do we collectively standardize towards something that everyone can use and ultimately teach machines to speak to machines with a person on the front and on the back?
00:07:17.091 --> 00:07:29.569
So for you, erin, and looking at the storage, why is this important to the storage practitioner and to the storage architect?
00:07:30.451 --> 00:07:32.699
Well, because it's coming, whether you want it to or not.
00:07:32.699 --> 00:07:33.084
Right?
00:07:33.084 --> 00:07:37.091
We thought ChatGPT and generative AI were disruptive.
00:07:37.091 --> 00:07:57.278
I feel as though this is accelerating at an even faster pace than generative was and and basically you know you will be faced with it, and so you might as well prepare yourself, you know, and maybe get some initial, you know considerations to think about.
00:07:57.278 --> 00:08:03.112
Right, because, on the one hand, it shows great promise, it can save a lot of work.
00:08:03.112 --> 00:08:15.475
On the other hand, it's so trendy right now that there's a possibility that some people will create something agentic that doesn't really need to be agentic.
00:08:15.475 --> 00:08:42.817
Right To me, there's this whole challenge around finding fit-for-purpose use cases and finding the right use cases, and I go into that in great detail in this session too, because it even took me a while in my role to find a use case that I thought was not just fit-for-purpose but also provided value tries to meet customers where they are in their journey, because, even though the hype is about running autonomously, the reality is.
00:08:42.937 --> 00:09:05.619
My belief is that most enterprise use cases will want a human in the loop for many of their actions, at least to start until they build that trust, and so it's going to be interesting to see how enterprises adapt to this, and my goal is to help people navigate through this disruptive technology before they're faced with it.
00:09:06.605 --> 00:09:15.470
Well, I think the core, what I love about the approach that you've got and I've had lucky, I got to preview some of the slides and see what content is coming, so it's really great.
00:09:15.470 --> 00:09:18.845
Some of the slides and see what content is coming, so it's really great.
00:09:18.845 --> 00:09:23.726
But, most importantly, as a consumer of this stuff, this is not meant to be a zero to hero.
00:09:23.726 --> 00:09:26.072
This is what I call it hype to hello world.
00:09:26.072 --> 00:09:32.874
This is where, like, we hear about this, but let's see it in its first form right.
00:09:32.894 --> 00:09:51.254
And I think this is such an important discussion because if we aren't doing the right requirements gathering up front and the right architectural decisions up front, we will pay for them 100-fold if we don't do the right things now.
00:09:51.254 --> 00:09:55.957
Different than all other compute before, especially in storage.
00:09:55.957 --> 00:10:00.461
This is it's a it's a wild west world, but it's.
00:10:00.461 --> 00:10:02.142
I'm excited by it.
00:10:02.562 --> 00:10:03.802
Definitely.
00:10:04.885 --> 00:10:36.542
Now, michael, just as I know we don't want to give it all away before we get to the webinar, but if you don't mind, from your side of the world, then representing SNEA and all the work you've seen around building standards, coming together as collective competitors, in effect, right, so, because there's participating vendors in the community as well as with IETF and IEEE, what is the beauty to you of what we are now doing as a community to get this information to the world and show how we are working?
00:10:36.542 --> 00:10:44.176
We're basically building in public and as competitive vendors that are cooperating with building in public.
00:10:44.176 --> 00:10:52.350
I feel like this is a great sample of what we want to see more of and what excites you about this webinar in particular, michael.
00:11:01.404 --> 00:11:02.427
Yeah, thank you so much for that question.
00:11:02.427 --> 00:11:04.754
So we have so many things going on within SNEA.
00:11:04.754 --> 00:11:05.475
There's multiple communities.
00:11:05.475 --> 00:11:06.759
Cst is only one of those.
00:11:06.759 --> 00:11:33.451
There's a lot of action behind the scenes that are focused on AI, and probably what excites me most is just the opportunity to work with experts in the field, like Aaron, and so there's so many folks that we're engaged with, and we'll bring that information through more webinars.
00:11:33.451 --> 00:11:38.071
And I did wanna say one word about what I'm most excited about.
00:11:38.071 --> 00:11:58.674
I don't wanna say too much about Erin's content, but she also has a demo, so this is going to be very exciting, because this really shows the tangible state of the technology that it's here now, and so this is something that's real right now.
00:11:58.674 --> 00:12:09.721
And so just the rest of the webinar, we just want to hold that for when we get to June 26, 10 am Pacific.
00:12:09.721 --> 00:12:12.462
So please join us for the webinar.
00:12:12.482 --> 00:12:12.705
Thank you.
00:12:12.705 --> 00:12:19.152
If you only see one webinar this year, you should probably see more webinars, but this is definitely the first one you should see.
00:12:19.152 --> 00:12:20.807
It absolutely is.
00:12:20.807 --> 00:12:26.639
You know, real experience being shared.
00:12:26.639 --> 00:12:31.966
This isn't a fireside chat with a bunch of AI people saying how they think it could go.
00:12:31.966 --> 00:12:57.181
I love that you've actually really identified proven, real market trends and thank you for that because, again, as somebody who's always looking for data that backs the pundit stories, it's so refreshing to see data-driven webinars with people like yourself, Aaron and Michael, who've actually done the hard yards and are working this stuff in production environments.
00:12:57.181 --> 00:13:00.014
So super exciting, Excellent.
00:13:00.014 --> 00:13:00.625
Well with that.
00:13:00.625 --> 00:13:08.620
So for folks that did want to connect with you before and even after the event, Aaron, what's the best way that they can reach you?
00:13:09.325 --> 00:13:11.552
They can connect with me on LinkedIn, for example.
00:13:12.595 --> 00:13:13.157
Perfect.
00:13:13.365 --> 00:13:15.152
Just search on Erin Farr and you'll find me.
00:13:16.905 --> 00:13:17.407
And Michael.
00:13:17.407 --> 00:13:21.455
Of course I'll say they should come to the next Nia SDC event.
00:13:21.455 --> 00:13:25.513
Whatever it happens to be, join the community, but there's obviously many other ways to reach you.
00:13:25.513 --> 00:13:27.412
What's the best way to get a hold of you?
00:13:31.365 --> 00:13:52.876
Yeah, so LinkedIn is fine, but also we have askcloudplugfest at sneaorg is probably the best way to get involved with our PlugFest and test tools activities, and I'm on that chain, um, so that'll be great.
00:13:52.876 --> 00:13:56.510
And you also mentioned uh, sdc, so that's in september.
00:13:56.510 --> 00:14:07.033
Uh, that's the 15th through the, uh, I believe the 17th, and uh, so we have three days of plugfest going on there.
00:14:07.033 --> 00:14:15.250
There's numerous presentations that are going on at the same time, so that's a really fantastic event Innovation in community.
00:14:15.465 --> 00:14:17.750
It just doesn't get better than that, so excellent.
00:14:17.750 --> 00:14:21.451
Well, thank you both, and so for folks, again, make sure you get signed up.
00:14:21.451 --> 00:14:24.474
We'll have links below and make sure to check out sneaorg.
00:14:24.474 --> 00:14:27.726
We'll have access to get registered.
00:14:27.726 --> 00:14:35.472
And if, for whatever reason, you're seeing this one and it's already passed, don't worry, because all the webinars, including all these great podcasts, are also available on demand.
00:14:35.472 --> 00:14:51.134
So, thank you, aaron and Michael, for your time today, good luck on delivering a fantastic webinar and demo, and I'm looking forward and I'll see you all on the live session and I'll also see you hopefully on some upcoming SNEA Experts on Data podcasts.
00:14:54.427 --> 00:14:54.688
Thank you.