Yasmeen Ahmad, Google | Google Cloud Next ’24

[Rebecca Knight]

Welcome and Introduction

Welcome back everyone, and welcome back to day two of theCUBE’s live coverage of Google Cloud Next here in beautiful Las Vegas, Nevada. I’m your host, Rebecca Knight, along with my two co-hosts, John Furrier and Rob Stretchay. Gentlemen, what a dizzying array of new product announcements, updates.

Pace of Innovation

The pace of innovation here is just mind-blowing.

[John Furrier]

Yeah, the data is the center of the action.

Data Analytics

Data analytics, our next guest from Google will break it down for us. Excited to dig into this one. Rob and I have been waiting for this session. Yes. Don’t be afraid.

[Rebecca Knight]

Introducing Yasmeen Ahmad

I want to welcome Yasmeen Ahmed to theCUBE. She is the managing director of data analytics at Google. Thanks so much for coming on the show. It’s a pleasure to be here and just sharing everything that we’re doing in data. Truly, the pace is just mind-blowing. Yes, so tell us more. I mean, as we said, there have been so many new announcements and updates. What is really standing out to you as we are here, day two of this conference, 30,000 people here?

[00:01:07]

[Yasmeen Ahmad]

Momentum with Customers

For me, it’s just amazing to see the momentum with our customers and what they’re doing with our technology and with all of the innovations. In the data analytics space, we have 40-plus new announcements that we are making over the course of yesterday and the next couple of days. It’s really the time where enterprise data and GenEI are coming together and changing the game for how business users experience data, how data teams work with data. Really game-changing in how the whole industry is being disrupted with GenEI.

[John Furrier]

Enterprise Truth and Gemini

One of the things I wanted to observe with you is that, okay, Thomas Kurian laid a lot out. He didn’t cover a lot of ground, but he said a couple things that were very interesting. Ground and enterprise truth. Yes. Okay, he also talked a lot about BigQuery, but you got the Gemini story is the big news. Okay, then you got now the Vertex integration at the Looker and BigQuery. So you got integration going on with Vertex and more Gemini. How’s that impacting the analytics products that you’re involved in?

[00:02:01]

Because that’s going to be the key enabler for the user experiences and the value that’s going to come out of that.

Impact of Gemini on Analytics

What’s the big change or what’s the big news?

[Yasmeen Ahmad]

GenEI and Enterprise Data

Absolutely. So two different areas of innovation I would talk about. First is bringing GenEI to enterprise data. For the customers we work with, they don’t want to take their data out of their secure data cloud. They want the BigQuery Vertex AI direct integration, meaning you’ve got access to large language models on top of multimodal data now. It’s not just structured, it’s images, documents, videos. So that direct connection between BigQuery and Vertex AI really enables enterprises to make use of GenEI on top of their enterprise data. And then the second category is the whole Gemini experience in BigQuery and Looker, providing assistive and agent-like capabilities. And you saw on the keynote, various agents, including the data agent, really it’s powering the users to be able to do more with data, but also automating parts of the data analytics lifecycle that has historically been super repetitive, mundane, and long and hard work.

[00:03:10]

[John Furrier]

Retrieval Augmented Generation (RAG)

You know, one of the biggest application successes we’ve seen in GenEI in developers and users is the RAG, where retrieval augmentation generation has been the talk of the industry. Because it’s easy to do if you’ve got data to work with. Not super, I mean, it’s kind of easy compared to other things. It’s easier than building machine learning algorithms. The vectors have been a huge support. You guys, that’s a big part of your announcements.

Vector Support

Why is everyone going crazy for vector support? And what does it mean to have the vector support in your portfolio versus other opportunities out there? It’s all the craze right now, the vector database and the vector embeds.

[Yasmeen Ahmad]

Yeah, so when you talk about retrieval augmented generation, being able to ground models, being able to tune models, you can do that now in BigQuery. And so with the various techniques, whether it’s RAG or doing LoRa adapters, it enables you to use your enterprise data to ground a model so it knows your business.

[00:04:06]

So that’s super powerful. And vectors are really taking off because today, enterprises typically have 90% of their data unstructured. It’s documents, it’s videos, it’s images. And historically, you could not search that data. You couldn’t really do anything with that data. But with Genii and vector embeddings now in BigQuery, with our unified data foundation, you’ve got the ability to create embeddings, do vector searches, compare documents. If you’re a HR team, you can match job descriptions to resumes. If you’re looking at various financial reports in document format, you can extract those financial metrics and compare them. So the possibilities now of unlocking that 90% of unstructured data in enterprise, that’s the major thing that is the attraction, right?

[Rob Stretchay]

Security and Governance

And how do you see this really evolving? Because you mentioned it, people don’t want to move the data around. Data has gravity.

[00:05:01]

How do you look at it from a security perspective as well? And how has that really been a talk track with your customers and how they’re interested in that?

[Yasmeen Ahmad]

Yeah, and I think there’s a lot of concern about security and also governance. Because in the world where you’re bringing in more AI and you’re doing more automation with AI, you need to know that the data on which it’s operating is trusted, it’s secured, and it’s governed. And with BigQuery, we are the AI-ready data foundation. So we are building the, it’s a unified foundation with one access control layer, one set of governance across structured, unstructured data. And actually with BigQuery Omni, it extends across clouds. Because we know many of the enterprises we work with are multi-cloud. And so being able to have that single unified access layer, single governance across that data, means that for our customers, they feel like they have a trusted foundation to deploy Genii on top of.

[John Furrier]

BigQuery Integration with Gemini

Some of the announcements, real quick clarification. I know there’s a GA of BigQuery integration with Gemini 1.0 Pro, that’s available now.

[00:06:03]

The rest are previews, are they public previews or describe the status of these projects. How do people get in on it? What’s the preview process?

[Yasmeen Ahmad]

Yeah, so our customers can sign up today for, for example, Gemini and BigQuery, Gemini and Looker. Those are some of the areas that are in preview today. This is where you’re seeing the assistive conversational analytics. You can now chat with your business data. This is where you’re seeing Data Canvas, which is this entirely new graphical workflow experience for data engineers to build data pipelines, for data scientists and analysts to explore data. All of these items, customers can sign up today and we’re very much looking forward to seeing our users get hands on with these technologies.

[John Furrier]

Just to be clear, public preview is a public preview, okay.

[Rebecca Knight]

Feedback on Conversational Analytics

Yes, what has the feedback been so far? Because I know we are here, people are getting their hands a little dirty with this, experimenting with it. What has the feedback been, particularly on the conversation side? Because I think that that’s a lot of skepticism with AI, is that it doesn’t feel real or human, it feels like a robot.

[00:07:06]

[Yasmeen Ahmad]

So I think on the conversational side, and in particular with enterprise data, what’s really important is you need the Gemini models, the large language models, need to be able to understand your business to give trusted answers. Because otherwise, it’s a generic model. So what we’re doing, for example, with Looker, is we have our semantic layer. And that semantic layer contains your business data definitions. And so we tune the models so that they have access to semantic business data definitions. They have access to metadata usage history. So that gives the model a grounding in your business. So when you’re doing conversational analytics, it actually comes back with answers that are very relevant. You don’t have to explain to the model what you mean by customer segments or target. It just gets it. So the feedback on that has been super, because context is going to be critical for really that accuracy, and for adoption by enterprises on that tech.

[00:08:04]

And building the trust, as you said.

Non-Data Engineers Using Looker

Absolutely.

[Rob Stretchay]

I was going to say, do you see a lot of organizations leaning into this because they’re looking for non-data engineers to be able to use Looker, non, you know, more of the business analyst side versus the data engineering side to go and create these outputs and understand the data. Is that really why we’re leaning into this?

[Yasmeen Ahmad]

It’s a huge, I think absolutely, and a huge reason why organizations are leaning in is data has been a hard discipline. It has, today it’s very mundane, it’s very repetitive. It takes a long time to pull data, build pipelines, get to insights. And typically the way we work today, it’s not very iterative, so it really slows organizations down. And if you just look out there at how many job adverts there are for data engineers, scientists, analysts, the industry doesn’t have enough talent.

[00:09:00]

There’s not enough people. So, you know, with Gemini and BigQuery, Gemini and Looker, it’s ways of accelerating and having more people in the organization and having access to data and being able to do it faster and quicker. But I also think there’s an interesting, we can have humans working on more of the creative task, more of the understanding and outcomes of the business instead of trying to wrangle these data pipelines. So for me, I think it’s, when I look at it, it’s not necessarily a, it’s eliminating rules or so, it’s actually creating time to focus on thinking about the outcomes for the business and how you get to value.

[John Furrier]

Getting Started with BigQuery

Ismael, I asked you about the BigQuery being the place where you can put the multi-models, you get the fine tuning and the grounding of the enterprise data. That’s compelling as a platform. How do people get started, one? And two, if I have a multi-diverse environment outside of Google, how does that interaction work? Obviously, you don’t want to move data around if you don’t have to. That’s preferred by most customers. Talk about that, the one, how do you get started with BigQuery?

[00:10:03]

And then two, if I have other data outside of Google, how does that work?

[Yasmeen Ahmad]

So with everything we’re doing with BigQuery, the unified platform from data to AI, we’re really bringing a much more seamless and easy for users to get started. What some of the feedback we heard historically was, Google, you’re innovating so fast, but we can’t adopt it all. It’s coming at us so quickly.

Unified Platform Stack

So rather than slowing down.

[John Furrier]

You don’t want to slow down. Keep going, faster.

[Yasmeen Ahmad]

But we’re bringing it all together in this unified platform stack. That means for a user getting started, we’re just GAing BigQuery Studio. There’s one entry point, and you get access to streaming data, whether it’s real-time, you’re doing machine learning, AI, all from that single environment. So that’s really making it much more easy for enterprises to get started. And to your point that today, data lives in lots of different places. So for us, we’re very much focused on being open and cross-clouds.

[00:11:00]

We know your data sits in AWS, Azure, many organizations of SaaS applications that are producing data in different clouds. So with BigQuery Omni, we’re connecting that data because we don’t think you have to move it all. And so it’s very easy to get started because it’s not the historical, you must move everything together and then start doing analytics. It’s leave your data where it is. You might be exploring, and at some point, you might move it, or you might not move it.

[John Furrier]

But the- If there’s value there, why not move it? Depends on the trade-off.

[Yasmeen Ahmad]

It depends on the trade-off, and it depends on the use case. And then with the seamless integration with our Vertex AI platform, you don’t have to be an AI engineer to take advantage of Gen AI. You’ve got access to Gemini Pro. You’re able to run really sophisticated large language model operations just from SQL. So it’s really empowering those everyday analyst users to make use of this technology.

[John Furrier]

Younger Generation Developers

I got to ask you, Rob and Savannah and Rebecca and I were talking yesterday about the next generation developers that are coming in. Like my son’s graduated from college this year.

[00:12:00]

He’s like, dad, that’s your cloud.

[Rebecca Knight]

Google’s my cloud. Okay, so- Not your father’s cloud.

[John Furrier]

So the younger generation, they’re also, they don’t have the bags yet. They’re like fresh thinkers. They just want more compute. So they’re coding away. So I can see this being attractive to younger developers, but they’re also thinking, I need horsepower. So that’s also coming up in the enterprise. I want to turn on performance. So I need to have performance. How does the analyst tap into that?

Performance and GPUs/TPUs

What’s the connection down to the GPUs and the TPUs? How do you spin up more power?

[Yasmeen Ahmad]

Yeah, so we’re super lucky. We have the stack. We’ve got the GPU, TPU technology. We have the AI platform. We’ve got the data platform. We have the business intelligence layer. So the way that Google is architected, because we’ve got access to all of that, we can really leverage that to the most. You know, we very much could integrate across those layers to get the most out of the entire platform. But we’re also serverless architecture.

[00:13:00]

The way that Google scales and how BigQuery was originally built, it was built for an internal use case at Google. It was built to run Google’s internal services. We built in a serverless way that means we can scale to super large degrees in very small increments, separating storage and compute. And that’s why enterprises are coming to Google because we have the scalability. We have the security. We have the performance. And now with the Gen AI integrated in, it’s up to your imagination what you want to do.

[John Furrier]

AI Agents and Workloads

Yeah, Rob and I were talking during the research meetings we’re having at our CUBE research team. AI is with the agents. We see a future where, and some movement now and some I won’t name names, but older applications, lift and shifting those workloads and putting AI agent around it to either manage it or in some cases projects where the person left the company who’s documented the code. So you’re seeing new use cases where go document everything.

[00:14:00]

So a lot of mundane tasks. And also lift and shift workloads into Google Cloud. So can you share any insight there how you see that evolving? Because we’re predicting that you’ll see Google take on new workloads that were running either on premise or in the other clouds and just wrapping AI around it for the lack of a better description.

[Yasmeen Ahmad]

Acceleration of Workloads to Cloud

Yeah, and we’re seeing a huge actually acceleration in workloads moving from on premises into the cloud. Because actually, if you want to leverage the benefits of Gen AI and all of the automation and agent like capabilities we spoke about, today that’s in the cloud. And so we’re seeing, we’ve actually seen an acceleration in customers looking to move legacy workloads, pipelines up into the cloud. And to your point, you don’t need humans managing them in the cloud. Cloud just works differently, first of all. You don’t need the traditional DBAs and so on. But also with Gen AI, you’ve got the assistive capabilities to set up operational monitoring of existing workloads.

[00:15:02]

You just need them to run. You’re not necessarily doing anything with them while you have space and time to do more innovation.

[John Furrier]

Consumption Side and Analytics Tools

What about the consumption side? Obviously, visualization’s always been hot. We’ve also seen data science, business intelligence groups. I won’t say losing power in the organization, but with AI becoming much more software, engineering, system oriented, the roles of business intelligence and data science teams are being augmented with AI. That’s changing the role of what applications are going to be consuming. So if I got all this data, I want to present it. I want to integrate it into the user experience. Can you share how that works with the analytics tools? How easy it is? What do people do?

[Yasmeen Ahmad]

Yeah, and we’re spending actually a lot of time doing, for example, integrations with our Looker Stack and Workspace. Because actually, the future of data is meeting the user where they are. Today, users have to come out of the slide or documents that they’re working in and go to their BI tool. Whereas the future is, I’m working in slides or I’m writing a report.

[00:16:01]

I want my assistant there and providing me the data or insights I need to plug into this report. So I do see that world infusing of where it’s not a separate data environment, it’s not a separate place to go. It’s actually meeting the user where they are and where they need those data or insights.

[Rob Stretchay]

Yeah, oh, sorry, go on.

[Rebecca Knight]

Gen AI Revolution and Next Steps

Well, I just wanted to ask, one of the things we hear a lot is that we are really still in the early innings of this Gen AI revolution. If 2022 was the year ChachiBT was released and you said 2023 was the year that people started searching for these proof of concepts, and now we’re here where it is being fully integrated into the enterprise, I’m curious what you think is next for this. I mean, where do we go from here now that it is fully integrated? When are we going to start seeing this creativity explosion and this great ROI for the enterprise?

[Yasmeen Ahmad]

And I think we’re already seeing it.

[00:17:00]

So some of the actually stellar customer examples we’ve seen here at Next, you’ve got customers in production, the likes of Puma, Priceline, they’ve pushed their first, second, third Gen AI applications now into production. And so it’s been incredible to see because it’s not just been the pace of innovation on the technology side, but actually the pace at which our customers have been able to adopt and actually push things into production, I’ve never seen it be so fast before. And I think part of the reason there is AI used to be the domain of the AI engineers who lived in a floor, and it was really hard to do it. You could say it, the nerds, yeah. We’re okay. We’re nerds, we’re proud of that.

[Rob Stretchay]

Yes, very proud.

[Yasmeen Ahmad]

We’re very proud. But with Gen AI and all of the innovation, it’s actually taken a discipline that was actually fairly mature, like AI models, et cetera, and made it now available to the masses.

[00:18:00]

So the ability, the technology has always been there. It’s actually just the unlock has been now opening that to the masses and making it, as you said, integrated and seamless to use. So where do I see us going from here? I think we’re going to see more and more production use cases coming out. And we’ve already got customers talking about all of the things that they’re now able to do and the efficiency that they’re getting that is creating the time and space to go after more ambitious use cases. And it’ll be exciting to see how industries change with this, right? It’s not just the internal. Right now, there’s a lot of internal, within organizations, driving efficiency. But once you get that internal organizational efficiency, what can those companies do externally? And what offerings or products or services, how do they evolve?

[Rebecca Knight]

Data Science Talent Shortage

Yeah, yeah. And then finally, I just want to return to something you’ve been talking about throughout this interview, which is the dearth of data science talent out there. As particularly as a woman in tech, and I’m a woman asking you kind of a sexist question, but how do we get more young people into this, and in particularly women, more into this industry?

[00:19:10]

[Yasmeen Ahmad]

Yeah. I mean, for me, this industry has never been more exciting. Like, if you wanted to be at any point in time in tech, like, now’s the time. The innovation, et cetera. I think earlier we touched on, you know, young people are gravitating. I think they are seeing the excitement of this space and what it can do. So I think we need to capitalize on that momentum. I think getting more women into tech, super, super important, and we don’t have enough females in tech. But all of the use cases and things coming out, and with generative AI and the creativity, that’s something that we can also tap into to get more people interested in this space.

Breaking Down the Nerd Misconception

You know, we joked about being nerds and being proud of it. It’s breaking down that misconception of what a nerd is. A nerd, it can be a really cool…

[John Furrier]

It’s mainstream. Nerds are cool. Nerds are cool.

[00:20:01]

[Rebecca Knight]

Absolutely, I mean…

[John Furrier]

That’s why we’re cool.

[Rebecca Knight]

Terrific to have you on the show. A really fun conversation. Thank you so much. Thank you, thank you. I’m Rebecca Knight for John Furrier and Rob Strachey. Stay tuned for more of theCUBE’s live coverage of Google Cloud Next. You are watching theCUBE, the leading source for enterprise news.