 Welcome back everyone to SuperCloud 4. I'm John Furrier, your host. The topic on this episode is generative AI, and it ranges from all impacts from society, government, technology, applications, but also the developers who are making the applications need to have the data. We have a great guest here who's going to comment on this, Jethna Mahajan, Chief Digital Information Officer. Thanks for coming on theCUBE, appreciate it. Thank you, thank you for having me. So we did a big feature story on developer productivity. Paul Gillan, a great story with you participating. The big, you're at amplitude, which is in the middle of all the action, cloud scale. So you see the cloud scale and the analytics and innovation right now is at all time high. Data is a big part of it, right? And so developers like security in other areas where that's not like their big thing, don't want to attend meetings and have to deal with like the schemas and which data warehouse do they use. So the developer role is going to be big. And you're seeing now with large language models and foundational models, there's an infrastructure developing, but not enough to use the analogy, picks and shovels available for developers. This is a big part of it. Meanwhile, the developer productivity conversation in the cloud native world is still at an all time high. So cloud native is all about developer productivity. And now with AI as the gift coming in, it's going to be very interesting to see. What's your thoughts? Yes, I think Gen AI, the two big personas that are really leveraging is the students and the engineers. So I think it's very, very disruptive. I'm very positive about it. I think it's just like the next, what same disruption as the, what the smartphones and internet brought in. I think it's as disruptive and I'm extremely positive about the Gen AI. And for developers, I agree. I think for the longest time, we've always struggled to get the software in time, the time to value. So with Gen AI coming in, there's tons of opportunities for developers and engineers. Their roles are changing. I think a lot of times they can do auto code generation, snippets, they can do testing. All the tech debt reduction is all time high, I think, with solutions like GitHub, co-pilot coming in, the time to value and putting software functionality. It's gone up really high. And I believe that this is revolutionary, although it isn't there, but I still think human oversight is extremely important. So having what I call responsible AI is very critical for developers to be able to use that. So yes, it's going to increase productivity, efficiency. As I said, it's gonna have auto testing. You could do a lot of prototyping ahead of time. So all that work is now gonna be shifted to Gen AI tools so that the developers can really think ahead. They can be the true problem solvers. They can design the solution to what the users need. They can actually think security and design user experience as they're developing. The co-pilot piece. Well, first of all, we've seen the three waves. The chatbots put that aside. That's clear, that's for the entry level stuff. The co-pilot or the human assistant has been a big game changer. And then the third layer we're seeing in the market is this predictive aspect of the data. So let's talk about co-piloting or human augmentation. Heavy lifting that used to be done, that can be done with AI or the advances in data architectures that are coming because now data feeds AI. If you don't have the data, but then compute and token limitations are now another factor. So you start to see the early days of, okay, we invented the wheel, we invented fire, we get what it is, now we've got to put into action with AI. What are some of the things that you see as use cases for developers getting the action here? Prototyping, co-generation? Yes, co-generation prototyping, as I mentioned, some of the use cases that are already there for engineers and developers. And there are two, some of the use cases you've already talked about, the customer engagement side of the house where chatbot embedding that. We've done that for amplitude as well, assisting, being the data assist for our product. We've generated that. So some of the other use cases that are around, making sure the data quality is there. I have seen use cases where developers are using to generate synthetic data, so they can really help expedite their testing. I've seen use cases where they are actually designing a solution through natural language. So as a requirement, you feed in requirements and they spit out some kind of a code and then you take it forward from there. So you don't have to do a lot of repetitive, menial kind of work that was traditionally done by developers so they can do the more sophisticated user experience kind of roles is what I'm seeing these days. So I've been talking on theCUBE a lot about and riffing on my podcast with Dave Vellante about there's going to be a creative culture that's going to emerge. Not that there hasn't been a creative class in society as always, in the nerd tech community. Creative has always been, oh yeah, UI or do this or architects are thinking differently. But if all the manual labor is going to be taken care of, say code reviews or any kind of automation that takes the humans' toil or grunt work or rock fetches, whatever word you want to use, it just makes them more productive. It gives them more time. With more time, they can focus on things that they like. This is going to, we think huge explosion of creativity. Huge explosion. And I think also there is going to be a lot of application proliferation because people are going to create. So there's a lot of risks around making sure that there is governance in place, there is infrastructure, all of that. So there's going to be a shift. I don't necessarily think that the work is going to go away but the shift is going to be more user centric, more data centric and more, I would believe it'll be security compliance by design. So those are the focus areas now for the engineers because even if they're making commercial software or they're using embedded AI in the existing enterprise application space, I think what will be critical is to enable the business, enable your end customers. So thinking that in those lines will be, I call that creative yes, but I also call it empathetic, lead by empathy. So I say deliver solutions that your users truly need. Thinking about things that the users may not even know. So when the smartphone came, they didn't know that we didn't know we needed it and now it's become something that we can't live without. So something similar along those lines is where I think. You mentioned, one of the use cases now, synthetic data, data generation, big startup area in Silicon Valley seeing right now and in New York and other areas around the world is this idea of data augmentation where you can't get access to the high quality data. You have to kind of simulate it or, what is synthetic data? Explain the concept because it comes up a lot. Is it like test data? Is it just something? I mean, what does it do? What is it, what does it do? Yeah, so for the, you know, from what I have through my experience, I've seen that for developers, the hardest or the testers, the hardest part is creating, generating data to test without compromising privacy, compliance, all the regulatory compliance that we have to go through. So we know we have to mask the data, you know, makeup data that takes forever for the analyst and the testers. So synthetic data would be where you train the model to be able to create similar, like the name, first name, last name, contact information, revenue, without it being the actual data. So you don't compromise on customer data, but you're feeding that like customer data that you can feed into the product so that they can test against it. All use cases, all edge cases can be covered. So you're creating data synthetically, but really not using the live data that compromises privacy and compliance. So in here, I want to talk a month because it gets into kind of like where this is going. Without the right data, AI doesn't work well. We also see compute's expensive, we just did a segment on that. And so you have also context window issues around token in, first token in, what are you optimized for through, but all these, I mean, a whole nother level of analytics is emerging for the developer. Even the notion of memory, it's not just physical memory, like in a systems context, but memory from a retrieval standpoint, if you had vector embeddings as an example. So this brings up the whole world of like, as a developer, these new analytics circuits, new observability kinds of things are emerging. What do you see as key markers for developers to look at as they put their toe in the water, look for a platform, look for the picks and shovels, the tools to help them build apps? That's right. So I really truly, that's so well said. I agree, I think developers, they haven't thought about analytics for their world. At the most observability would probably be the closest that they come to. Now they can do, they can run insights around, if I like optimizing the code, the error handling, the different options that are available to them, they can pick and choose to design conversions, like what will it look like from one programming language to the other, orchestration of what comes first and next in terms of the order of execution. They can also, as you said, compute understanding like what that does into the performance of their code that they are going to generate. So those, and then also reusing, reusability of snippets of code and then they can optimize a lot of that design, reaching out to the libraries. Cause a lot of times, you know, there's a lot of, when the developers develop things, they often forget about where it is in the library because sometimes end up rebuilding all of that. So having access to those libraries, being able to use AI generation, AI tools to be able to get that information to your doorsteps, it's like amazing for them to be able to drive time to value. Productivity angle is interesting because it's a whole another wave and still early people like to use innings, we love innings as a sport, but you know, Amazon CEO calls it the three steps in a 10K race. I don't even think the game has started yet, but you look at the data, we're just talking about like to the old school recommendation engines. You can't just add new stuff. You have to rebuild everything from scratch. Now you can kind of stream stuff in. Don't have to do batch. You got distributed concepts of data sets. You got lake houses, which are nice environments for serverless and other, you know, abstracting computer way versus the old school data warehouses. So the world is changing. The old data warehouse world, databases notions are up in the air. What's the big trend paradigm? That's the biggest script flip or paradigm that's going to upset the apple card, if you will, from the old school, because you have old way and now you have this new way. What's the most important thing happening? I think the most important thing that in my mind is actionable insights. So earlier there were insights and then we had to kind of figure it out what actions we need to take. So it's a step further in the maturity level, the way the data has been presented. However, I would say that a lot of risks are there. There are biases, there is hallucination that we talk about. So we need to really take into account the risk. Oftentimes ourselves included because of security concerns, we usually turn off our configurations where we're saying, okay, we're going to use external data that's available to us, the white labeled data, but we're not going to train, let the model train on our data because then that's proprietary and stuff like that. So how do you- Or the ways could be different? Yes. How do you create the balance between that and making sure that it's- Another thing I would really emphasize is as it's expanding and growing, the creating the governance and center of excellence and that framework to be able to allow it to do all of that and the balance between external data and our internal data, I think that is going to be extremely critical in that phase. The role of data is key. We see hallucinations, everyone sees that on ChatGPT and OpenAI. All the time. Obviously it's proprietary large, of course. Generic prompts create hallucinations, but if you prompt engineer properly and now create prompt merges or code, API calls, it's starting to see an art and science. So we see a trend where, I'll call it an AI wrapper as an app. That's not a bad thing. I used to be like, ah, it's not a big deal, but actually it could be compelling. If you're just a developer, you want to wrap around OpenAI because it could be cheaper to use their inference, for instance, then try to build your own. And that reminds me of the old website days, because you just build a website on top of the web. And the web is now OpenAI or Anthropic or one of the large proprietary language models. That's not a bad thing. I totally agree. So I personally think, even for the developers, their skill sets are going to change. There's going to be a lot of education there where they are going to figure out more orchestration between different AI solutions that are there. As I said, they're figuring out when to use what human oversights were kind of testing they need to do on top of what's been generated through AI. So I really think that that's the area that they're going to focus more on. That's where the bank for the buck would be for investing into our developers and engineers. And as you said, like the use cases that are there, at least for that we are using in our company, we've used Cognitive Search. Within the company, we've used GitHub Co-Pilot. We're using a lot of, like when I talk about Cognitive Search, we've done knowledge management around that. So I think education around that would be critical. Yeah, so as you said, creating the development around AI skill sets would also be different. It's going to be developing on top of what you already have. An area that's coming up is integration with development environments. Yes. Okay, so I'm a coder, I got my IDE, and all of a sudden I have the context switched into another environment, may have different policies and rules. This actually was probably an untouchable area from a human standpoint, because it's like the indifference, I mean, too much work. Yes. I can't do that, or I don't have the training. That is going to be, yes. Cross coding, this is a really interesting part. Huge, yeah. So I think with the GNI coming into picture, the concept of ID is just going to expand so now you could do natural language conversion to code generation, so that a lot of junior developers that coming in, they will have huge assistance in being able to develop that. You know, now using, being used to a particular IDE or an environment, that'll go away because you can use the GNI and tools to be able to do that translation and bridging for you, for the developers. So I think it's a great space to be in, and that's a great opportunity. Like in the content market that we're in, we're in the media, we generate content. Dave Vellante had a great quote on our Q-Pod. He said, chat GPT makes a good writer great and a great writer exceptional. Exceptional. And developers, almost the same thing you mentioned, junior developers, because, you know, I was just joking last night on the phone with some startup founders and we were, he's got a platform engineering, AI platform, and I was riffing on the whole Mark Andreessen 10X developer. Back when the cloud, he said, hey, you get one developer does the work of 10, kind of the old school, full stack, waterfall based developer, then you got cloud or agile, iterate, you know, the whole, you know, lean startup kind of mentality. Now you get crafts coming back in here. So you have two factors going on. I want to get your reaction to one is, we've been saying there's a 1010X goes to 100 or 1000X developer because AI can make that developer even greater. But also this creative class brings craftsmanship to the opportunity where old school software was very crafty, slow, we had to do a lot of QA, shelf wear days, remember those days. So now we're seeing a lot more craft. In fact, I was just talking with the CEO of Chrome, Chroma, and he's so proud that they have such a great developer uptake that they like his product. That's a very craft mindset of crafting good for the personas you're trying to sell to or build to. Talk about that. Yeah, so I think in terms of now that all the grunt work, if you will, in the coding world, QA, testing, tech debt, all of that is going to be taken care of with tools like GNI tools. I think the focus is going to be more on the experiences. So when I say craft, I feel like it's more around experiences. Wherever the digital touch points are, mobile experience for the app, your experience on the web for the functionalities. So thinking ahead of the consumers and the users. And then there is going to be almost instant conversion of the feedback that you get from the customers, whether in product or through the community. And you'll be able to have that insight, the external instant insight from your consumers and the consumer behavior to be able to start developing the product roadmap. So I think that problem solving, the experience craft is going to really boom in this area. So really focused on delivering what the customers need and faster and being able to think ahead of the... It's going to change the startup certainly and go to markets because you can't... I mean it's going to be really whoever does the best job will win, the better mousetrap will be there. Exactly, I totally agree. So I think that's why I'm thinking about some of the GNI tools that I'm seeing. Oh my God, like I'll give you my example. As a CDIO, we do a lot of access controls for all our enterprise software or including our own commercial software. So we're seeing GNI tools that are coming in that is going to automate who's accessing, what time, live. And I'm like, oh my God, this used to take months and now it's literally in front of our screens one thing. So it's very, everything is very instant in life and it's really helping with, as I said, the experiences, the security posture and all of that. So I think it's going to be... It's a step function change to get to the application use case you want to get to that required, old school manual labor building blocks or undifferentiated heavy lifting or toil, whatever word you want to use, that's really where the game changes. Yes, and what another thing is, it brings in the external world insights to us. Like, okay, you know what, this is what your competitors are doing or this is where the world is going. So bringing in those insights, which would take us a little longer to see once our product is on the market, seeing how the responses and things like that. Now all that is getting accelerated. So as you mentioned, it's becoming more of a craft. It's a great, great conversation. Certainly we'll do a lot more conversations with you and really bring you in, unpack a lot of these topics, we're just scratching the surface. But it is SuperCloud here and SuperCloud 4, which NAI, a lot of things we're touching upon is developer drill of experience. One of the things that's clear in the radar is the notion of personalization. Because if you get the data and you have the chatbots and you got the co-pilot assistants and you got predictive, the next level up is going to be personalized world. Absolutely right. So there I have an opinion, I would share, is so it's extremely important to do it right. So jumping the gun on this kind of personalized customized experience, I believe, is sometimes can hurt the company with commercial software. So extremely important to do it right and do it slow. Low hanging fruits, things that have already been there, like chat, JPD and stuff like that. Or engagement through chatbots and things that's extremely beneficial. And I agree, and having a screen and personalized, even for sales people, like we're starting thinking, okay, when a rep goes into, sales force is doing, when they go into their software, they are starting to figure out what's best for them. They bring in all the information, so the 360 degrees on their account to be able to make informed decisions ahead of time. So that is extremely critical to making sure that when you think of personalization and customization, you do it right. Otherwise, it'll get frustration into customers and then they'll form an alliance. It's not personalized, it doesn't work. It doesn't work, yes. It's like saying, hey, why don't you send a personal to you? Yeah, you don't know me well enough. It's like, wow, you didn't read the room. Yes, exactly. So to speak. It's a blind spot. Agreed. Well, we appreciate it. What's your role at Amplitude? Put the commercial in for Amplitude. Really appreciate you contributing to our program here. What are you guys looking to do? What are some of the things that you're working on that you're excited about? Sure, so I'm the chief digital and information officer at Amplitude, so my role is making sure we drive data-driven culture, eating our own dog food. So Amplitude is, we build a platform and we power people's products so that they can build exceptional experiences for their customers and provide product-related insights. So my role is to be able to drive that employee experience, building the corporate technology so we can have our enabler business to run their company, including a product organization and engineering organization, so they can deliver the best-in-class product and services for our customers. Congratulations on your success. We've been following your company since you guys were on our startup showcase, our Amazon web service, the startup showcase thing with season one, episode one, you guys might have been on. And now public company, congratulations. Thank you so much. Thanks for coming on. This is SuperCloud 4, Genevieve AI, the impact to the industry, mainly as we're driven by developers, a tsunami of new developers and a generational shift is happening. You're going to see developers looking for those picks and shovels, so to speak, to build those apps, the infrastructure, it's all going to change and it is legit next level. We're covering it here on theCUBE. We'll be right back with more after this short break.