411: Snowflake オブザーバビリティ部門GM ジェレミー・バートン――データ層はコモディティではなく、勝負の本丸だ

A few queries run on your enterprise data will give you vital context that you can supply to the LLM so that you can burn less tokens over here. And so I feel like in the enterprise, at least, if you want to get the return on your AI investment, you've got to do some work on your data foundation. Otherwise, you're not going to get the context and actually not get the right answer, or you're going to blow a lot of money getting to the right answer. So it's going to be one of those two things. It's the future. Don't hit the snooze. AI is you're giving humans the boost. Team humans rolling. Brains in the game. With Dan Turchin, y'all remember the name. People ain't in the mix. No stopping the float. Better work, better life. Yeah, we're ready to go. Robots and humans are collab, you know. Okay, let's start the show. Good morning, good afternoon, or good evening, depending on the way you're listening. Welcome back to AI and the Future of Work, episode 411. and I'm your host, Dan Turchin, CEO of PeopleRain, the AI platform for IT and HR employee service.

As always, if you've enjoyed this great conversation or any other, you owe it to humanity to pay it forward. Share a link to this one or any other that you liked with just one friend and they will know that you care about making humans better with great technology. If you leave a comment, I just may share in an upcoming episode like this one. Thank you to Tony from Austin, Texas. Hook them horns, whose favorite episode is number 159 from back in 2021 with a shoe guard, general partner at Foundation Capital about finding and investing in the future of AI. That was a fun one. The shoe's been in the news quite a bit. He's been talking a lot about context awareness for AI. Go look that one up. We learn from AI Thought Leaders Weekly. And of course, the added bonus, you get one AI fun fact each week. Here it is for today. Anthropic co-founder Chris Ola was invited to speak at the Vatican in response to Pope Leo XIV's encyclical Magnifica Humanitas as part of Anthropic's initiative to widen the AI conversation in his remarks.

Ola summarized Anthropic's perspectives this way. AI made displaced human labor on a massive scale, making the support of displaced workers a moral imperative. Because AI development is concentrated in a few wealthy nations, we need a mechanism to ensure its economic gains are shared globally. As AI systems become embedded in everyday life, society faces pressing questions about what it means for individuals, families, and communities to truly flourish. The internal structures of frontier AI models are increasingly mysterious and unsettling. Researchers are discovering internal states that functionally mirror human neurological results and emotions. It's a pretty dense read, as is the encyclical. Of course, I'll share a link to that in the show notes. My commentary on this one? The Pope's encyclical echoes themes shared by Pope Leo XIII in 1891. Following the first Industrial Revolution, Pope Leo XIII condemned both unrestrained capitalism and revolutionary socialism.

Similarly, Pope Leo XIV rightfully says AI is not morally neutral. Both embraced the need to not reject modernity while also not letting technology violate human dignity. Just as Pope Leo XIII didn't slow the pace of innovation, Pope Leo XIV's encyclical will only cause us to ask more thoughtful questions, which in itself is an important outcome. I applaud the Pope for using his platform and influence to share a message that humans can and must coexist with machines. Intelligence will never be outsourced to bots. Only humans have true agency. Because we dictate what AI can and can't do. It's a pretty dense fun fact for this week. but go and read up on that and we'll continue to unpack some of these topics in future weeks. Now shifting to this week's amazing conversation. Jeremy Burton's the GM for observability at Snowflake. Post-acquisition of his company Observe, where he was the CEO. Prior to Observe, he was EDP of Marketing and Corporate Development at Dell,

where he also served as President of Products for EMC. Jeremy joined EMC from Serena Software, where he was CEO. Prior to that, he held leadership roles at Symantec, Veritas, and Oracle. He's also a long-serving member of the board of directors at Snowflake and a member of the McLaren Racing Advisory Board. He received his bachelor's in engineering from the University of Surrey, southwest of London, which, of course, served as the training base for the London 2012 Olympic and Paralympic Games. One thing that Jeremy has not done is appeared on this podcast that's about to change. Jeremy, welcome to AI and the Future of Work. Let's get started by having you share a bit more about that background and how you got into space. Yeah, thanks for the intro. I appreciate it. No, look, I mean, as the accent would suggest, I am not a native of these parts. Although I've been here 30 years. so i say tomato and aluminum and uh all the good words you all americans will understand um but no

i i actually came out to the u.s in in 95 um long time ago 20 30 years to work for a guy called mark benioff um mark was only a couple years older than me at the time uh we were both a lot younger and and far more handsome back then. But Mark was working on a new product, and I came out to learn about it. And I ended up – I told my wife I was staying two weeks, and I ended up staying six weeks and did the launch demo with Mark on stage at the Moscone Center and with Larry Ellison in attendance. And I moved out six weeks later. So, yeah, I've been here a while. Always done enterprise software. Most recently, a startup. uh i my entire career had done big company things and i really badly wanted to do something from scratch so when i joined observe it was you know sort of a handful of founding engineers um and it was as much to prove something to myself as anything which is that i'd worked

a lot of big and successful companies but you know really that that bigness and successfulness was not really down to me. I helped, for sure. So then the question is, well, can you get in at ground zero and make a difference and get something going and off the ground? And I think those are the folks who probably deserve most of the credit for getting these huge corporations to what they are today. So, yeah, a little bit more meat on the bone. Incredible journey. Great story about Benioff back in 1995. you made a fateful decision when you were starting Observe to hitch Observe to Snowflake. When I think about the observability space, whether it's Datadog or Dynatrace or others, I think about applications first, not about data platforms. What insight did you have that led you to team up with Snowflake early on? Yes, a couple of factors. is, you know, when I left Dell, I mean, EMC had acquired the company I worked at.

Sorry, Dell had acquired the company EMC that I worked at. I really wanted to go do something that was like in the core of data. And I'd been on the board of Snowflake at that point for probably three years. And I was like, okay, I thought the database world was over. I thought it was going to be reinvented by Hadoop we all probably can still remember Big Data and Hadoop and then when I came across Snowflake I was like ah this is really like a re-architecture and a reinvention of the database and I felt you know probably after seeing it grow for a couple of years that it was going to do great and that there would be companies that would be sort of snowflake shops. And, you know, 20 years earlier when I was at Oracle, it's like history repeating itself, right?

I saw that firsthand. Anyone who built an application on top of Oracle, a good one, actually did quite well because certain companies were like, yeah, we're an Oracle shop. Do you want to put more data in Oracle? You know, great. Like, bring it on. and so I was looking to do something which was like looking for a company that was building on top of Snowflake and I talked to three or four of them sort of Hill Ventures that were one of the very early investors in Snowflake they had funded a number of companies that were building on top of Snowflake and the observed pitch resonated with me not just because they were building on Snowflake but if you look at the problem of observability and it's still somewhat the same today as it was seven years ago, if you want to see if there's a problem with your system, if something's broken, you would look at a metric and you would have a time series database and a tool that would sit on top of it. But if you wanted to dig in and investigate, oh, well, then you don't, that's not metrics, that's logs.

And by the way, that's not in a time series database, that's in this bespoke system over here we call Splunk. and then if you wanted to go dig into latency performance oh no, that's actually not logs you've got to look at spans and traces and that's in a system over here and it's like a new relic or an app dynamics and so the sort of big brain thought was quite simple, it was like why don't we put that data in one database and then when we see a spike in a chart we can see the logs for the spike and we can see the spans for it. And so the founders, the thing that resonated with me was like observability is really a data problem. I know people are saying that they've got lots of tools and it's confusing and it's expensive, but they have lots of tools and it's expensive because they have fragmented data. And so Snowflake was really one of the first databases to come along and elegantly deal with sort of the diversity of data, sort of unstructured, semi-structured, structured data, as well as the scale.

And it was really like a new database architecture that would enable them to scale. And we had a couple of early Snowflake engineers on the Observe team as well. So that helped the decision along quite nicely Philip Unterbrunner who did the query engine and Vadim Antenal who did a lot of the unstructured data support in Snowflake It was quite a controversial decision. We lost a chunk of the engineering team over it because there were plenty of folks in that early team that wanted to build their own database, which was not a stupid thought because, you know, I mean, Splunk built their own database. Honeycomb built their own database but we felt like hey if we didn't need to build our own database we could spend more of our engineering time adding value on top of that database and ultimately that's what people are charging for so yeah we lost a bunch of the team we didn't know at the time we made that decision whether we could actually build an interactive

observability experience on top of what was inherently back then a data warehouse which typically is batch mode type processing. But we convinced ourselves after six months it was possible and never looked back. It was prescient because you foreshadowed an AI era when there's so many more applications, so much more software, and most importantly for your business, so much more telemetry data and observability is infinitely harder. Yeah, I mean, you know, back when we started, a terabyte a day, five terabytes, that was a lot. I would tell you we have mid-sized companies, let's say 1,000 to 2,000 employees, that have got 30 to 50 terabytes a day at this point, and large enterprises that are 500 to a petabyte of data a day. So the data volumes are quite astronomical. And just to put that in perspective, when I started at EMC 15 years ago,

we had one customer in the whole world that had a petabyte of data. Since the beginning of time to 2010, they'd collected a petabyte of data. We have some customers that will collect that every two days. Unbelievable. You have had a catbird seat through your board role and now obviously through Observe on a really exciting epic journey. at Snowflake. I had the pleasure of working under Frank Slootman when he was ServiceNow. You presumably did when he was EMC. He famously left ServiceNow to lead Snowflake. Now Shredar is the CEO there. And by the way, on this podcast, we recently had Arnon Kishuri, who is the chief people officer at Snowflake. Could you give us a little bit of an insider's tour of the culture at Snowflake? And I know it's a big factor in why Snowflake has been so wildly successful. yeah i mean look i go back to sort of the two founders um benoit and and thierry um i i i knew

i knew both from oracle i knew thierry better i actually played soccer with with thierry he he was a far better engineer than he was soccer player which is which is which is great right because um so so they um there's a like the engineering culture i'd say in the company is incredibly strong um it's a very sort of warm accepting almost collegiate type place um you know the product has been wildly successful i think you know when frank came in you know his power alley is is sort of scaling from a few hundred million to to to sort of low billions and i think you know probably then the very sort of early engineering centric culture um sort of broaden to include the sales team you know when you've got to start to land like some of the biggest and and largest accounts in the world then you know you've got to build up more muscle on the go-to-market side of the house um and then look uh i think shridhar now coming in um

it has actually been i obviously i knew shudhar from from from from being on the board and i think the big challenge of not just snowflake but any company the size of snowflake is um there's a new world out there there's a new reality around ai um you know snowflake and any other company you know created before the last couple of years is not an ai native company and so how do you get a company that's 10 000 people to to think that way and i think shreda just deserves massive massive credit every bit as much credit as maybe Frank deserves for scaling the go-to-market side of the business for really getting the company on the same page around AI and what it means to be an AI-native company and it's painful. There's no easy way to do it. Observe, I ran a company of a couple of hundred people. It's quite easy to get a relatively new company onto a new trend

because you don't have many people and, you know, by nature have been in a startup. I think you are much more attuned to the latest trends as they come along. So, yeah, I think, you know, Shredar is different to Frank, is different to Bob Muglia who was the original CEO. But I do believe in sort of a leader for a time and Frank was certainly the right guy to bring in. I mean, Bob was the right guy to bring in initially. You thought of your engineer's engineer and deserves a lot of credit for innovating, like the pricing model and all kinds of wonderful things. But Frank, to take the company public and scale it, and I think for the AI era, you've got Shridhar in. And as I said, I've been amazed, actually, by just the depth to which AI is perpetrating every pocket of the organization. It is something to behold. I had a chance to walk the halls of campus at Snowflake with Arnon recently. in menlo park and it there's an energy about it um that i hadn't seen yeah maybe i think a pair

analogies since the early days of service now there there was really a really good morale it felt like you know everyone was there to do great work um it must be different though to be acquired in after being at observe and it's certainly you know on the sidelines of snowflake but what's the transition been like yeah we we are probably quite fortunate i think in that um snowflake said doing well as a separate company, observability is a different business to sort of traditional AI data cloud, pipelines, data warehousing, data engineering, all that kind of stuff. So why don't you keep running the thing over here? You keep going to the software engineering leaders, CTOs and so on, selling observability, and then let's figure out where the leverage points are. I mean, Snowflake does have great relationships and so many big accounts. it's not that their relationships maybe are responsible for observability, but they've got good access to the people who are.

So the nice thing is I am still able to run Observe. I was a big believer early on, and I've always felt this way, that when you join a bigger organization, yes, you're going to have your own traditions within your team, and if you like that, that creates your culture. But you can't deny that you're part of something bigger and better And so I was quite insistent that day one, on the close, we moved to the Snowflake offices in Menlo Park and be part of the team. I didn't sort of want this us and them mindset. It's like, okay, we're in this to make a difference for Snowflake. Let's be part of Team Snowflake. But, of course, we can retain our traditions and things that we have done as a small company to sort of like mold the culture. Now, the AI tooling companies would say, including the foundation labs, would say, we own the cognition and the data layer is a commodity. And presumably Snowflake and maybe your competitors would say,

your cognition is dumb without the context provided by our data. Where do you come out on this? Who owns the value from the perspective of the user and the enterprise when nobody wants to cede any ground to anyone else. Yeah, it's an entertaining debate. I think in the business that we're in observability, when you think about hundreds of terabytes a day, okay, if you just want to point an LLM at your hundreds of terabytes of telemetry and have it got figured out, like, good luck. It's just not happening. And so what we do is not just manage the volume of telemetry and sort of put it in a big dumpster for you to dive into, we actually do a lot of work on the data to structure it, to organize it, and to relate it. And that context, particularly when you've got hundreds of terabytes of data a day,

that context is critical. And whereas I think that the frontier models, clearly they can pick up knowledge around Kubernetes or infrastructure, it's all on the internet and those models are trained on the internet. And so, yeah, it can probably do a decent fist of providing good context around Kubernetes troubleshooting. But if you're a bank and you have a mobile banking application, okay, the knowledge about how that works is not on the public internet. The models are not trained on it, right? We have the data from that or those kinds of applications. we understand the environment and it is that environment that is the key context for the agent in the troubleshooting flow and I think what's happened sort of thus far not surprisingly is hey we've got LLMs, we've got agents let's point them at this, this, or this and let's see what they can do and I think what we're finding is that

we can burn a lot of tokens very easily and we can blow our budget very, very easily and I do think the thinking has got to sort of come back or that the pendulum has got to come back towards the center, which is, by the way, if you provide more context to the LLM, you will burn less tokens because you're going to get it more of a clue of what to do next or where to go look next. And so my view of this is we've got to sort of balance the CPU burn with the GPU burn. A few queries run on your enterprise data will give you vital context that you can supply to the LLM so that you can burn less tokens over here. And so I feel like in the enterprise at least if you want to get the return on your AI investment you got to do some work on your data foundation Otherwise you are not going to get the context and actually not get the right answer,

or you're going to blow a lot of money getting to the right answer. So it's going to be one of those two things. It's clear that the snowflake strategy in kind of an AI-first world is very ambitious, far more than what most snowflake watchers have come to expect. Give me the bull case on what AI services am I going to Snowflake for that I wouldn't traditionally think to go to Snowflake for. Yeah, I mean, I think fundamentally the data foundation has never been more important. And it's maybe not surprising to hear. If you're in the frontier model business, it's always easy to sort of look down your nose and say, well, yeah, all of that stuff beneath us is a commodity. I mean, I've been doing this a long time. Managing massive volumes of data, particularly critical business data, never has been a commodity, and I don't believe will be a commodity now or at any time in the near future.

I think what folks are also realizing is that if I have 10 source of truth in my enterprise, which almost every enterprise does, an agent does not reconcile those across those sources of truth. It might average across them, but it will give you the wrong answer in a confident way. And I think what enterprise is realizing is, okay, but if I can reconcile that source of truth, right, and if I can provide some semantics around that data, then the agent can do it work and it will be game-changing because, you know, what we've always wanted is sort of, you know, I think right back to, you know, Microsoft in my formative years in the mid-90s, you want information at your fingertips. And that has been an elusive dream, I think, until now. But you only get enterprise information at your fingertips if you do some legwork on the data foundation. I would also tell you, for the longest time, you know, Snowflake in the early days,

it was all about just putting the data inside Snowflake and reasoning about data inside Snowflake. I think what is interesting right now is Snowflake quite early jumped on Iceberg. I think most large enterprises are going to look to iceberg as sort of the open table format. And so what that means is that the catalog, you know, Snowflake, you know, introduced that probably a year or so ago, becomes fundamentally important. Because if you've got the vast quantity of data in iceberg format, some of it may have been written by Snowflake, some of it may have been written by something else. you need to know like hey where is the data that I need to query to do my thing and the catalog I think starts to play you know quite a critical role and then I think more broadly you know Snowflake is I think aggressively adopting

coding agents I mean Coco you know is their coding agent and you think well okay we're not using it to sort of you know generate Python code to build applications so much, but if you want to set up pipelines, the data infrastructure required to lay down that data foundation, I think the sense is it's never been easier to do it. We can take all the legwork out of that, which is quite involved and hairy, and no one values that much the plumbing, but you've got to do it. It's like I've got lots of plumbing in my house. I spend a lot of money doing that plumbing, but actually nobody sees it. And I think that that was the feeling around, you know, any data foundation, which is I've got to do a lot of legwork and I've got to migrate data. The promise, I think, is that AI and things like Cocoa are just going to make this very, very easy. And there's quite a massive library of skills that Snowflake has produced around, you know, sort of data engineering and data operations

just to make life much easier. And I think that is now starting to show up in the business, right? It's much easier and faster to do things. It's much easier and faster to ask questions. That drives the query volume, which is really what drives Snowblakes business. That's what I thought you would say. I was hoping you were going to give Coco as the example. So thanks for sharing that. Today's sponsor is a product that I've used and have loved for well over a decade. I encourage all the companies that we partner with to use it as well. It's Box, the leading content platform for the AI era. It acts as the secure, essential content platform where AI agents not only access unique institutional knowledge, but they can orchestrate end-to-end workflows across business-critical systems, delivering the kind of enterprise AI transformation that you expect. It's the key to unlocking the power of AI. As we all know, LLMs are pretty naive. They don't know about the specific content stored in files across your company.

And without access to that rich content that lives in product roadmaps and sales materials, research, HR policies, financial models, and other content, there's no way that they're going to be able to orchestrate workflows in the way you expect. Fox is building the intelligent content management platform for the AI era. It's the secure bridge between enterprise content and the AI that you want to deliver to your business and your customers. Go to box.com slash AI to learn more. Again, that's box.com slash AI. Now back to this week's amazing episode. I think it's pretty easy to make the conceptual leap from where we are today with AI primarily based on language modeling to kind of the frontier, which is AI entering the physical world. What's the impact on Observe and Snowflake when the consumer of AI, let's say it's a rocket or, I don't know, it could be an HVAC system,

or it's certainly a non-human and it's generating different telemetry data and we're thinking of videos and images and a whole vastly different set of things, artifacts that need to be monitored, stored, analyzed. How does that change your business? Yeah, I think obviously the volume goes up, right? And I think with the advent of Kubernetes probably 10, 15 years ago, is it really that long? You saw probably a 3 to 5x growth in telemetry volumes. I firmly believe you're going to see that again in the AI era. I think then the volumes of data then are obviously so vast. to ask a poor old human to help you figure that all out and find root cause when the rocket explodes or there's a problem. No, you're going to need AI to help troubleshoot the data that has been generated quite often by AI.

So needless to say, we think that this process of structuring and organizing the messy machine-generated data, AI can help with that. I mean, if I go back, I mean, observe one of the sort of unique innovations that we produced was something called a knowledge graph, and we did this back in 2018. And the knowledge graph really is just, it's an abstraction that sits on top of this machine-generated mess. But we originally designed it to enable humans to navigate that data much more quickly. Instead of diving into a dumpster of logs and looking for a customer ID, you could say, oh, look, here's all my customers. Here's the sessions. Here's the pods they were running on. Here's the containers. So it turns out that we didn't know it at the time. It turns out that we were actually total geniuses because it is that structuring and that knowledge graph that we use to give context to the agent.

And so, you know, wind the clock forward now. Okay, yes, agents can use that knowledge graph, but what if we could use an agent to build that knowledge graph? Today when we build it, you know, if you had an application, we would sit down with you and say, oh, when you have troubleshooting, what kind of questions do you ask? Because every business is a little bit different. You know, our customer, Dalpied, looks at calls. top golf looks at golf bays these are all different entities that businesses want to ask questions around and we you know we would have our our implementation guy sit down with the sre team and say you know and ask these kind of questions and then we would extend this knowledge graph to better reflect their business but you start to think okay what what is the art of the possible now. We could just ask those questions. We could analyze the query results. The agent could suggest structures or entities extracted from that data. And then as we further ran queries,

it could optimize the query performance. So I think AI agents will essentially end up creating that graph. That graph then becomes the basis for other agents to drive the investigation flow. so you can start to see, yes, there's more telemetry flowing in the bottom. The good news is we don't have humans creating entities and structure, and we have agents doing that. Oh, and by the way, that structure they've created is going to help other agents advance the troubleshooting flow more quickly. So, yes, there's going to be humans involved. There's unknown problems you've never seen before that will need some sort of human insight, But I think the goal here is that 90% plus of the legwork here is really done by a collection of AI agents. Makes sense. In a pre-AI world, as much as this sounds like an insult, a lot of observability could be performed using heuristics. Back in the day, we didn't think of that as an insult.

That was just, you know, it was a nice, reliable way to observe applications. Now, as we introduce AI, kind of these probabilistic or non-deterministic outcomes, there's a different level of responsibility. It's like observe such a snowflake as a vendor. It's less about just verifying the integrity of the data or its state and more about taking responsibility for some of these actions the decisions that are made How do you think about the shifting role of what you do with respect to these probabilistic outcomes It a great question because I think for the last decade, there's a unit of currency that we've been trying to troubleshoot really is a microservice. A microservice is a lump of code that typically talks to another microservice that talks to a database. An agent, I know we profess that like this is something wildly different, but what's an agent?

Well, it's kind of a lump of code that will talk to maybe another agent, and it'll talk to a database, but it'll also talk to an LLM. And so that changes the picture quite a lot because I always describe an LLM as like a new animal in the zoo. We know databases. We've had them for years. We know how to monitor them. We know how to investigate them. it's a well-known problem domain um llms we don't know that there isn't the level of observability if you like into the llm that we would get from say like a database um also i think uh you touched on that a second ago a database gives you sort of empirical facts back um it's a deterministic system you you will get the same answer given the same question you'll get the same answer every day that is not true of an LLM and so I think what becomes more and more critical in these as we try and troubleshoot these agentic systems is a way to evaluate the LLM

there's going to be certain bounds within which an answer is acceptable and we have to constantly run these evals against those LLMs to see if the answer is still within the bounds of acceptability and if it's not that you're like what why is that where something obviously has changed. It could be something with the data source. It could be something with the way it's been prompted. There's many possible reasons, but I think what we would call LLM observability, in particular evals, is just going to be a critical part of, I guess now we'd call it the harness for the LLM. What's one of the essential things that you need in order to run an LLM? It's going to be evals in order to determine how the responses are changing over time. And by the way, that could be down to a new model bin release. It's kind of topical given Anthropics' new release this week. That release changed dramatically probably the function of a lot of existing applications.

I think this is one of the big challenges with AI right now. Not only are we trying to go at a million miles an hour, We're trying to do it at a million miles an hour while the sand is shifting under our feet because the models themselves are still advancing quite dramatically. So there could be a time when observability owns evals. Yes, I think it's a critical component for sure. Each week on this podcast, we talk about the future of work. And one of the things I'd like to get your perspective on as a leader is this tension that we all face right now. There's obviously a limitless amount of work to be done. The amount of work to be done isn't going away. It's not shrinking. It's only growing. And yet there's this trope about reduce the size of our organizations, prevent bloat, dot, dot, dot, without editorializing. I would just love to get your perspective on, you know, as a leader, growing an organization, super ambitious.

Does the team, how does the team look different in one, two, you know, maybe three years down the line? Yeah, I think you've always got, with new technology, you've always got to re-examine, like, what is work? What work for, like, an employee? if there are clearly things that can now be automated and that was a large part of someone's job then I guess the role has changed and there's no, to me that's been the entire history of technology I mean 100 years ago, like 80% of the US was agriculture and farmers and then the tractor came along and we're like oh my god, we're doomed and it turns out we were okay And I feel like, yeah, the LLM is sort of the modern-day tractor. And right now, I think what's very hard for us humans to see past is what new work will be created.

I'm generally an optimist on most things. I mean look I loved all the coverage recently around the almost trip to the moon or the trip around the moon but it's like okay we did this guys like how many years ago? Two years after I was born and I'm really old and it's like okay what problems can we now conceive because we're not dealing with minutia if you think about I mean supersonic flight travel right we did that in the early 70s right we don't have a supersonic jetliner today so there's a part of me thinks like okay we can only conceive that a job that did x y and z today well x and y is taking care of about AI therefore you know we don't need that job anymore but if you look at the you take a step back and say okay what does this technology enable if we're operating up a level of abstraction what problems can we now reason about that we could never reason about before what things can

we now achieve that we could never achieve before what work can we now do that we could never do before it's i think it's impossible for us to to conceive of that until it sort of hits us between the eyes you know um it was impossible to conceive of like the hoover dam you know when we just had chuttles you know that that you know so you know i i feel like right now what's healthy as a leader is you have to examine the work that your team is doing you have to examine whether that can be done by ai uh the reason is if you don't examine it your competitor is examining it and they're going to be able to move faster and do it for less and blah blah blah than than you um and then it's also the job of a leader to think about, okay, given that we're moving 5X as fast, what can we now add to the roadmap that we couldn't do before? You know, I am still quite bullish on the case for software engineering and software engineers.

I just think we'll be able to take on wild projects that we couldn't ever conceive of. I do think education will have to change, right? I feel like all of these great things that were proven to be not that useful in software engineering back when I did computer science are actually more relevant than ever. Things like architecture, design, you know, like good practices around architecture and design will allow you to define a much better system for AI to go off and implement. but I think the last 40 years, you know, we've tended to dive straight into implementation and generally hack away or rewrite and probably not employ great software engineering practices, right? The nice thing now is implementation is promising to be free. That doesn't mean to say software engineering is free. You know, you can still mess it up with code generation.

Brilliant answer. Everyone listening, when you need a dose of positivity about what's ahead for humans, go ahead and replay that one. It is the case that we're going to have more time to dream bigger dreams, and I think that's net positive for humanity. Hey, Jeremy, about out of time. I've got to get you off the hot seat, but you're not going anywhere without answering one last important question for me. So rewind the clock back to the kid from Surrey getting off the plane, getting ready to meet Mark Benioff for the first time. What's your advice to that kid? What do you wish that you had known then that you know now? You know, a career is a long time. And I think when you're young, ambitious, you can't really comprehend, like, whatever, 30 years in the tech industry and you're in a hurry. and I think particularly in Silicon Valley,

like your superpower is your network. I actually wish I'd kept in touch with more people. I've met folks over the years who've been very, very good at this because all those folks that, you know, you kick around with when you first start your career, those folks become, you know, leaders and quite often huge companies. I mean, Mark has done quite well at Salesforce. The product manager I worked with for Oracle 8 was Thomas Kurian, who has a decent job over at Google Cloud. And so, yeah, I think you feel like you can do it all yourself, I think, when you're young. It's like sort of get out my way, I'm here. But actually, the superpower ends up in your network and who you can work with. And a lot of that is who you keep in touch with and who you build relationships with. So, yeah, I wish I'd probably valued that much earlier in my career. But, yeah, it was the best decision I ever made to actually get on that plane.

I will say that as well. Hateful. Well, good. Good advice. Jeremy, work for the audience to learn more about you and the great work that you and your team are doing. No, so we are social media is probably the best on LinkedIn, observing. I think we're just moving our website under Snowflake, snowflake.com slash observability. I'm sure it's going to be. But, yeah, follow us on LinkedIn. Follow us via the blog, actually. That's a good place to find the latest and greatest. Brilliant. Well, as the world changes, we'll have you back, and we'll see how some of these big ideas that you pose are faring in the real world. This is great. Thanks for hanging out. All right. Thank you. Thanks for the invite. Yeah. Well, that's all the time we have for this week on AI and the future of work. As always, I'm your host, Dan Turchin from Keeper Rain. And, of course, we're back next week with another fascinating guest. Yo, it's the end of the show. Time to power it down.

AI's the business for humans. Wear the crown. Team human in the house. Got the vibe so clever. Dan Turchin from Keeper Rain. Making us better. We learned about AI for humans. We're ready to conquer the future. That's all for now. See you next time