Ron Gabrisko joined Databricks at less than $1M in revenue and helped build it to a $7B+ run-rate over the next decade.
He’s the rare CRO who’s scaled a business from early startup to near-public company, runs his team on the product he sells, and has lots of advice around early sales motions and building and scaling a sales team.
This episode of The Peel takes us inside Databricks’ 13-year journey from a small startup founded by seven PhDs to a $190B company.
We hit on all their biggest decisions, the mistakes, what they got right, and what they wish they knew earlier.
If you’re building a go-to-market engine and want the playbook from someone who took it from zero to $7B+, or just want an inside view of how the biggest enterprises are adopting AI, this is for you.
Support this Episode’s Sponsors
Numeral: The end-to-end platform for sales tax and compliance.
Flex: Sign-up for Flex Elite with code TURNER, get $1,000 here.
Amplitude: AI analytics, all you have to do is ask.
Merge: Every model. One API. Total control. Check out Merge’s Agent Handler.
Monaco: The revenue engine for startups.
Inquire about sponsoring future episodes here.
Timestamps to jump in:
0:00 From under $1M to $7B+ in revenue
1:07 Seven founders and three big bets
2:49 Why going cloud-only was contrarian
5:14 Monetizing open source: "what will they pay for?"
11:40 What Databricks actually is
14:17 Genie, the AI he runs the business on
19:20 It's the data context, not the model
21:36 The early AI bet, before LLM's
26:32 Why enterprise AI beats consumer AI
30:00 Automating his own sales org
32:18 How Ben Horowitz pitched him
33:48 Why seven co-founders is an advantage
35:54 Teaching the CEO sales: org charts and MEDDIC
42:08 Biggest sales mistakes and four growth stages
45:14 Why technical products need technical sellers
47:21 The seller profile: technical, gritty, no short stints
50:45 Back-channeling references that don't BS you
53:32 Hiring 40 reps and why PLG didn't convert
58:07 How a16z opened enterprise doors
1:05:47 Why he gives POC's away for free
1:08:50 Raising prices to match value
1:11:34 Why he killed seat-based pricing
1:14:37 Build for enterprise requirements early
1:16:46 Consumption selling and the six-month planning cycle
1:19:54 Expanding internationally without breaking it
1:23:38 The four C's of enterprise AI
1:26:54 Why messy data blocks AI adoption
1:29:07 Forward deployed engineers: what makes them win
1:31:56 When does Databricks go public?
1:33:36 LL Cool J, Michael Jordan, and never losing a game
Referenced:
Careers at Databricks
Databricks’ Genie
Apache Spark
Relentless: From Good to Great to Unstoppable on Amazon
Find Ron on LinkedIn
👉 Stream on YouTube, Spotify, and Apple
Transcript
Find transcripts of all prior episodes here.
Turner Novak:
Ron, welcome to the show.
Ron Gabrisko:
Thanks for having me. Excited to be here.
Turner Novak:
I think this will be a fun conversation. You joined Databricks before you guys were really out of the woods, when it was less than a million in revenue. As of the time we’re speaking, the public disclosure is $6.9 billion in revenue, and by the time we publish this, I think it’s going to be even higher.
So this will be pretty fun, going deep on how you did it and how you got there.
Ron Gabrisko:
Yeah, it’s been an absolutely amazing journey. We’re still growing super fast. We say it’s top of the second inning, still early days. It’s an amazing business, and it’s been a great journey.
Turner Novak:
I love those baseball analogies. When you’re on an earnings call and the CEO goes super deep, it’s like, we’re still getting in the seats, the game hasn’t even started. We haven’t even thrown the first pitch yet.
Ron Gabrisko:
Well, I’m a baseball player. Early days, that’s what I wanted to be, a Major League Baseball player.
Turner Novak:
Oh, really?
Ron Gabrisko:
Yeah. Now I sell software. I guess it’s the closest thing I could get to the big leagues.
Turner Novak:
What do you think is the key to success for Databricks? If you had to sum it up in a couple of sentences, how did you go from zero to $168 billion? You can correct me, but...
Ron Gabrisko:
A hundred and...
Turner Novak:
Eighty-eight, I mean, it’s like a $188 billion company, in about thirteen years. Is that the number?
Ron Gabrisko:
Ten years. The company’s been around thirteen, I think. I’ve been here ten and a half years, and the company was less than a million when I joined. Most of that growth has been over those ten years.
The keys to success at Databricks? I can talk about how we built it in each stage, but overall, we made some big bets. First of all, we have seven of the greatest, smartest founders on the planet, all PhDs from Berkeley. They were at the frontier of data and AI way before AI was even cool. This is back in 2014, 2015.
They made some really strategic bets early on. One was to go all in on cloud, one was to go all in on open source, and the last was to go all in on data and AI. And that wasn’t really the vogue back then.
Since then, we’ve had the best engineering and innovation machine. Our product is the best I’ve ever sold. It’s the best on the planet. And I think we have the best go-to-market team on the planet. To achieve that kind of growth at this kind of scale is fairly unprecedented.
So the combination of those two things, plus great people and great culture, that’s been the key to success.
Turner Novak:
I want to talk more about all of those things. Going in order, when you talk about making a bet on the cloud, that seems obvious today. Why was it such a big deal fifteen years ago?
Ron Gabrisko:
Back then, cloud was still pretty early. Most of the infrastructure was still on-prem. When I started, I took all the founders out to Wall Street. I said, if we’re going to make a business here, we’ve got to sell to all the big financial services banks.
We met with most of them, with CIOs. I got the CIO meetings, and it was crazy. They would bring hundreds of people to meet the inventors of Spark. Matei would be signing autographs.
Turner Novak:
That’s awesome.
Ron Gabrisko:
But they were like, we’re never going to the cloud, and we were cloud only. They said, literally never. Now, some of our biggest customers, a bunch of these big banks, they’re all in on the cloud. It’s already happened. But back then, that wasn’t a no-brainer.
Turner Novak:
That’s kind of wild, because it would be like someone telling you today, we’re not going to use AI, we think it’s fake or it doesn’t work. What was the justification for not moving to the cloud? Was it not valuable enough, or too hard, or too expensive?
Ron Gabrisko:
Early in any market, you’ve got to overcome legacy. There were a lot of people saying the clouds aren’t as secure as our own dedicated security team, which obviously is false. These clouds have to have better security than any one individual company.
So a lot of it was security, governance, regulations. A lot of the banks are in regulated markets. It’s similar to the early days of AI. Once you solve those challenges around security, governance, scalability, and regulations, the market takes off.
It was a great bet. Everything’s moving to the cloud, and we could move faster than a lot of the on-prem players because they were stuck dealing with legacy issues.
Turner Novak:
And with open source versus closed source, today open source is not that crazy. Was it a little crazier at the time? Was it just not really proven yet?
Ron Gabrisko:
Back then, what were the big open source projects? Maybe Hadoop and Linux, and the model for how you monetized it was just support and services. So one of the ways Databricks changed the market was how you monetize open source.
We built a managed service. We built one of the first managed cloud services for open source. A lot of the hyperscalers were using open source to monetize their compute, but not a lot of private companies were doing that back then.
As we built out this model, we kept open source at the core everywhere. Now we’ve had many projects beyond Spark, like Delta and MLflow, on and on. Building a managed service around open source, and figuring out how to monetize compute usage, that’s the future. But it was a fairly new concept back then, and I think we developed it.
Turner Novak:
So at the time, you had a bunch of people using the open source product, and you started to get people to pay. How did you make that transition?
Ron Gabrisko:
Exactly. There were literally millions of people using Spark. Spark was the open source product early days. That’s why I loved the company as an opportunity.
First days, we just said, go meet with as many customers as possible who were using Spark, and understand how they’re using it, what their challenges are, and what they would pay for. It wasn’t rocket science.
Then you look for the trends. People will pay for security. People will pay for scalability. So you start adding these features as a paid layer on top of the open source. And we’re selling a managed service too.
Early days, we were selling to a bunch of the Silicon Valley digital-native startups. With open source, they tend to love building their own stuff. But when you go to the enterprise, they don’t necessarily have the expertise to build large open source projects, so they need partners. They need a managed service. So breaking into enterprise was a really big step in how we scaled the company.
Turner Novak:
When you say managed service, that means you do some of the work for the customer. You’re building some of the features, or updating things for them. How would you describe that for someone who’s never heard the term?
Ron Gabrisko:
All the features are there, right out of the box. You’re not having to set up each product, each piece, each feature, and configure it. It’s super configurable, but if you were to try to build something like that from scratch, you’d have to get 20 or 30 different open source pieces of software. You’d have to stand them up on whichever cloud you’re going to use. You’d have to configure all those things, and worry about security, scalability, and governance.
Databricks just comes ready to use, which was huge. We kind of created the data science market. Early days, I said we were doing AI before AI was cool. People were starting to do machine learning and data science, so we went straight to the data scientists. They were like, this is awesome, I can just upload my datasets and start working. I don’t need to do a bunch of setup with IT.
Obviously that evolved into everything we see today around data engineering, data pipelines, AI, and machine learning. But creating an easy-to-use managed service on top of open source was a big tailwind for us.
Turner Novak:
So how do you describe the Databricks product for someone who’s never come across it before? Honestly, this would be helpful for me, because I’ve heard of it all the time, I don’t really use it, and I’m not even sure I actually know what it is. What is the product for someone who doesn’t know?
Ron Gabrisko:
The easiest way to describe it is, we take massive amounts of data from many different sources and let you do AI, predictions, and analytics on it.
An example: you use a streaming service, Netflix or Disney. When you see the next best movie and it says it has a 98% probability for you, what they’ve done is taken millions of data points from other users, and all the other data you’ve given them, like movies you’ve watched or liked, and used that to predict your next best movie.
We apply that to pretty much every industry. We find credit card fraud for banks. How do they cross-sell and upsell customers? How do you do loan approvals, loan origination, R&D for pharma? Early days, I would say someone will solve cancer with Databricks. Lots of companies are doing that in pharma.
So the problem set is, how can you build models, do predictions, do recommendations, and build agents on data? And the more data you can feed those systems, the more accurate the predictions, models, and agents.
Even in today’s world of AI, everybody uses ChatGPT or your favorite LLM, Gemini, what have you. For an enterprise, how do you make decisions? It’s all about having the context of your data and attaching that to AI. So we developed a product called Genie.
I actually have it on my phone. I use it to run our business. It has all the context of my business, and I can just ask it questions in English. It does all the calculations, all the SQL queries, all the tech in the background. If I ask it, what are the top ten customers that might churn in Germany, it’ll build the churn model, try to understand the churn, and make recommendations on how to fix it. It’s pretty cool. That’s a high-level overview of Databricks. Hopefully that helps.
Turner Novak:
I was going to ask you about Genie, because you said you run your team on Databricks. So it’s the user-facing, simple-to-use interface.
Ron Gabrisko:
We run all of Databricks on Databricks. I can predict our revenue within one or two percent. I can predict which customers are going to churn, and which products are the stickiest, like customer retention.
Genie lets me do all the research and questions. It’ll prep me for a customer meeting. It’ll tell me everything it knows about the account, which products they use, and their opportunities to grow revenue, because it’ll do other research too. The interface is similar to any LLM. You just start asking it questions. The difference is it’s doing it on your data, versus generic, publicly available data.
And it’s able to do calculations, graphs, predictions, and launch agents. So it’s super sophisticated for enterprise AI, versus the generic tools out there today.
Turner Novak:
Has that always been there, or is it more of a newer product?
Ron Gabrisko:
It’s brand new. We launched it at Data and AI Summit. Now we literally have millions of users on it. You have CEOs of banks, CFOs of banks. It’s perfect for running your business.
Traditionally, how do you run your business? You have some dashboard or analytics tool. Maybe it’s updating in real time, maybe it’s batch. It’s definitely not doing predictions for you. It’s definitely not letting you just ask questions with all the context, not just from whatever that dashboard was built on, but the context of all your data.
It builds out the ontology of all your data sources, so it knows where to look for answers to certain questions. And it does calculations, predictions, machine learning models, and queries under the hood. So it’s the first of its kind, and it’s state of the art. It’s pretty awesome.
Turner Novak:
So instead of having the guy on the team who you’re pinging to say, hey, can you give me the updated numbers, you’re just messaging Genie, the Databricks product.
Ron Gabrisko:
It’s real time. What everybody has is an analyst. You say, can you go get me this answer? Then it takes a day, and you say, no, change this, add this. Then it’s a little stale. Can you get this week’s data versus last week’s? With Genie it’s all real time, right at your fingertips.
We want every business user to have it. We have huge retailers where the store managers say, how do I maximize revenue on this shelf or with this product, or which promotion should I run? That’s going to depend on where your store is located and your clientele. Rental car agencies say, how do I maximize sales? It’ll give you feedback, like, your CSAT is low on this, you should clean your cars better.
One of the big gas station retailers said, I want to maximize sales of pizza. They’d have a camera feed, and it would actually measure the pizza and pizza pricing. It puts the power at the fingertips of every single business owner. That’s the future.
Turner Novak:
That’s pretty interesting. In this pizza pricing example, I’m assuming it’s a security camera they already had in the store. What kind of data does it take in? And how does the pricing work? Is it real-time pricing, like, hey, it looks like we can charge an extra 30 cents?
Ron Gabrisko:
I’m sure it’s not like gas pricing, where it’s changing while you go to grab a slice. Like, oh, it’s now $15. I’m pretty sure it’s not like that. Grab your pizza fast, it’s going up in a minute.
Turner Novak:
Yeah.
Ron Gabrisko:
I think it’s more like, if they run out of pizza, they want to make sure they have pizza there. So it’s monitoring that. It could take video images. It’ll also figure out how many pizzas you should have that day, managing inventory based on seasonality or other promotions they’re running.
That’s a small example. The idea is, how do I maximize my entire business? A lot of these gas stations, it’s about petroleum sales, but a big part is their retail business, all the things they’re selling inside the store. How can you maximize the sales of all those things? That’s going to depend on location, promotions, and other factors and data you’ll have access to.
Turner Novak:
In terms of Databricks versus the market. I see it all the time, every day, though maybe it’s tailed off a bit. There was a time when every day there was a new AI product that solved all your business problems. There are a lot of them out there. So what do you think Databricks differentiates on? I’m assuming this is a pretty standard question you get from customers.
Ron Gabrisko:
There are all kinds of new models out there. Obviously OpenAI and Anthropic have their models, but now there are lots of open source models. I’m sure you’ve read about them, Kimi, GLM, what have you.
It’s not really about the strength of the model. The key for Databricks, and the key for a lot of these companies to be successful at AI, is all about the context of your data. The models are already smart enough. I ask these models all kinds of questions, and they’re as smart as I am, if not smarter. It’s all about the context you provide.
So your ability to connect those models with your proprietary data and your business context is really what unlocks the insights and the outcomes for these companies. That’s been the key for Databricks, in how we’ve built out our data platform. We were AI first from the beginning. Ten years ago, we were thinking about AI and machine learning before anyone else.
We approached the market from that perspective, in how we govern data, with Unity Catalog and Unity AI Gateway. We serve any model. Unity Catalog doesn’t just categorize your data, but all your models, your notebooks, everything else you’re doing, so we can automatically understand where your data is and what’s relevant to answer a question or build a model around it. That’s the key to Databricks. No other company has that breadth and expertise.
Turner Novak:
Back when you made this big bet on AI, it was probably not quite as obvious. What was the thinking? How much of it was that you got lucky with how AI evolved, and how much was you seeing where it was going?
Ron Gabrisko:
Part of it was just insightfulness from the founders, in the first products and use cases they built. Spark was a massive data processing engine, which meant the more data we could process, the better we could perform versus other products.
We went after the data science market first. A lot of these data science projects, think about a huge genome study trying to find a new cancer treatment. They’re taking a bunch of compounds and mapping them against different genomes and against electronic medical records, trying to find trends, like which compounds they should test to accelerate R&D.
These genome files were huge. Traditionally you’d say, let’s run one model to predict something, and it might take 24 hours. Well, Databricks comes along, and now I can run that in a minute. Think about how fast I can accelerate my virtual R&D.
That’s now reality for every business. How do I take billions, if not trillions, of data points, analyze them, get insights, and make decisions for my business? That’s been the key, versus being unable to do that in the past. So it was insightful, and certainly fortuitous, to start in that part of the market, developing machine learning and data science. That’s how we came at the market and entered AI early.
Turner Novak:
And LLMs were not really a thing back then, or I don’t know how good they were.
Ron Gabrisko:
No, they weren’t a thing back then.
Turner Novak:
So what was the initial product like? What was it powered by?
Ron Gabrisko:
We had notebooks. We still have notebooks. Data scientists are building machine learning models, and it’s all code. It’s not, hey, I can just use plain English to build these things.
Then you had the ChatGPT moment, when ChatGPT finally got good enough and consumers started using it. We actually developed our own model back in the early days of that, and it was pretty good. Then we bought a company called Mosaic, probably three years ago, to start doing training for custom models.
Our AI R&D team is as good as any on the planet. But we’re really focused on how to solve enterprise AI problems and outcomes for specific verticals and businesses, versus the rest of the market, which is more focused on general intelligence for consumers. That’s a big difference at Databricks. That’s our secret sauce, how you connect the AI to the data in the best, most secure, and most governed way, to unlock insights, build agents, and automate processes.
Turner Novak:
It’s been pretty interesting how, back when ChatGPT launched, there was a lot of excitement around consumer AI, and rightfully so. ChatGPT grew super fast. But when you got deep on what it was, it was a lot of people using it as Google, using it as a therapist, tons of questions, and you pay thirty bucks a month. You might actually lose money on a lot of those customers.
Versus on the enterprise side, a lot of the time it’s, hey, I’m doing this to complete work. On the extreme end, Uber paid a billion dollars in a couple of months for their enterprise AI usage. So the opportunity in enterprise over the past couple of years has been kind of insane, in how useful it’s been and how much work you can get done with it.
Ron Gabrisko:
It’s massive. I think we’re still in the early days. Everybody thinks we’re automating everything. Right now, we’ve solved one use case front to end, which is how you better automate coding. Claude Code, Codex, Cursor, all those guys are going after how you automate coding and accelerate software development. The tools are really good for that, and we serve all of those through our Unity AI Gateway.
But how do you automate all these other processes? How do you automate finance? How do you automate a lot of the admin work in sales? We do a lot of that automation with Genie, Genie Code, and Genie agents, inside our business and inside our customers.
That’s still pretty early days, because you have to understand the process and re-engineer it. You’re not just lifting and shifting the same process. But there’s tons of upside. We’re at very early days on how we do that, and that’s massive upside for this market.
Turner Novak:
Is it starting to tip in some of these other categories? Are you seeing practical usage, where you’re actually able to use it the same way engineers are using the coding agents?
Ron Gabrisko:
Yeah. We’re customer zero for this. We’re an AI company, so we’re going to automate and use AI in our entire business. And we bring those learnings to our customers.
A lot of this is about getting in there and having the expertise, because you have to redesign the process. You can’t just automate the same process. Look at healthcare. How do you automate claims authorization, claims payment, and finding claims fraud? You need to understand the process, teach the agents what to look for, make sure the agents are high quality, and then continually make sure they’re doing a good job.
We’re still at early days on how we automate all those processes, but I’m starting to see the early wins at a bunch of Databricks customers. That’s super exciting.
Turner Novak:
Internally at Databricks, you’ve probably added billions in revenue since the ChatGPT moment. How has using AI changed how you operate the team? Any tactical things someone listening could take away and do too?
Ron Gabrisko:
Totally. I run our entire business on Databricks. I can get super deep on any one customer. Genie will prep me for all my customer meetings. We’re starting to automate a lot of the sales sequences. Anything I can do to automate admin work, like forecasting and how people enter things into Salesforce. We’re a consumption business, so we call them use cases, but anything I can automate to increase productivity and sales time, I’m going to automate.
A hundred percent of my sellers have Genie on their phones. They can run their accounts and their business. They can do demos for customers. I’ll go to customer dinners and do a demo right there.
Turner Novak:
Just pick up the phone and show them.
Ron Gabrisko:
Yeah, I just show them. I’ll have them upload some dummy datasets for their business, and say, here’s how it can find opportunities in how you’re handling claims or loan origination.
We’re going to put AI in everything we do. It’ll diagnose your account and say, here’s the next best action. It’ll know, oh, you have a retailer, you’ve sold them this use case, you should go sell them this other use case, and you should talk to this person about this type of app. So it makes recommendations for all our salespeople.
They can do their own demos. Genie is super easy to use, so it’s great to demo for pretty much any user of data or AI. We’re putting it in our entire business. I’d recommend that for everybody, because that’s where the market’s going. It’s been great for us.
Turner Novak:
One of the most interesting things about your story with Databricks is that you originally joined after a conversation with an investor in the company. How did that conversation go, and what happened?
Ron Gabrisko:
I met Ben Horowitz from a16z. The guy’s a legend. He said, hey, I have this little company. It’s probably smaller than any company you’re looking at. It’s called Databricks, but it’s got the most upside of any company in my whole portfolio. Maybe you can meet these guys, or recommend somebody.
I met the founders and just hit it off. These guys are smarter than anybody on the planet. Their brains work in a different way, and they understood the space. I thought data and AI was the future, so I decided to take a big swing at it.
Ben was super funny. He said, these guys are Berkeley, they invented this great piece of software, the greatest piece of software on the planet, and gave it away for free. I need somebody to come in and help them build a business out of it. It’s a fun story. I’ve developed a great friendship with Ben over the years. He’s a legendary investor, and it’s been awesome for our company.
Turner Novak:
I have a question from one of my friends, Paul Klein. He’s the founder of a company called Browserbase, and he’s a solo founder. His question is, you’ve got to ask him, how do you put up with seven co-founders? How do you even do that?
Ron Gabrisko:
I get that question a bunch, especially from CROs. They say, I can only handle one founder, how do you handle seven co-founders? I think it’s a massive advantage, actually. You have seven true owners of the business.
Companies say, can I get a founder to speak at my event, or can I meet with a founder? They’re all super technical. If you’re talking to a technical audience, I have seven times as many people to do exec alignment with CTOs and CDOs.
Early days, engineers are generally skeptical of salespeople, so I spent a lot of time getting to know them, understanding their strategy, and how they wanted to develop the company and the culture. That would be my recommendation. When Ali took over as CEO, the first thing he did was sit down and say, teach me sales, I want to learn sales. We formed a great partnership building the company from early days.
So dig in. The founders are the company, especially early days. They set the tone, the culture, everything. It’s been a huge advantage with these seven guys. Love them to death.
Turner Novak:
Doesn’t Anthropic have seven co-founders too? Maybe that’s the lucky number.
Ron Gabrisko:
I have no idea. Maybe.
Turner Novak:
I think there are seven or eight, I can’t remember.
Ron Gabrisko:
That’s a good number.
Turner Novak:
You said something interesting. Ali sat you down and said, teach me sales. How does that conversation usually go when a founder asks you that? What do you walk them through? What do they usually struggle with? If I’m super smart, how do I learn more about selling things to people?
Ron Gabrisko:
First off, most early-stage engineers think, if I just develop the best product and have the best pricing, everybody’s going to buy it. It doesn’t really work that way. The power of sales, especially for enterprise, is that salespeople teach my customers how to use the product, how to get value out of it, and make them aware of it.
I started on a whiteboard talking about organizations. When you’re selling to startups, there might be one person you need to sell to, the CTO or the founder. But when you sell to enterprises, they make decisions as an organization. You need to understand the org chart, who the players are, who the decision makers are, who owns the budget, and how decisions and approvals get made. So we’d talk it through. Now Ali will ask, who’s the decision maker, who’s our exec champion, all those things about how you develop relationships.
Turner Novak:
Do you typically just ask people that when you’re first meeting them, or is that too tacky? How do you figure that stuff out if you’re meeting someone for the first time?
Ron Gabrisko:
No, I wouldn’t do that the first time. A lot of these sales cycles take six to twelve months or more. Initially you want to get to know a person and understand them. It’s about asking a lot of questions. Sales, to me, is more about asking smart questions and listening, versus having a glitzy pitch.
It was funny. When I was interviewing for the company, they said, do the pitch. I said, okay, send me the pitch. I looked at it, and it was super technical.
Turner Novak:
People probably just zone out. Like, I don’t feel like looking at this flowchart with all these diagrams.
Ron Gabrisko:
It had a lot of acronyms, a lot of speeds and feeds. So I had some advice on how I’d pitch Databricks, but I also started with, okay, what kind of company are you? How do you use data? How do you use AI? What other systems do you use? What are some objectives you want to get out of using AI with data?
And they said, are you going to pitch us? I said, I am. Because the more information I can gather on your challenges and how my product can help, the more credible I’m going to be as a salesperson. It’s like going to a doctor. If they just say, yeah, you need surgery, you say, don’t you want to know what’s wrong with me first?
So sales is a lot about coming in prepared, knowing about the customer, doing as much research as you can on their challenges and strategic objectives, going in high in the org if you can, and asking a lot of questions. Get to know them, understand how you can help, and then be credible, diligent, and trustworthy in how you follow up. Don’t bug them, but if there are specific areas you can help with, that’s what a good salesperson does.
Turner Novak:
You kind of have to assume, the bigger the company, the longer it’s going to take to build that trust and get them to move their company over onto your product. It’s a long process.
Ron Gabrisko:
Yeah. There’s a lot of technical validation in there, doing POCs and so on. But start with understanding the problem. What is the business problem they’re trying to solve? How can you help them get there faster, cheaper, and with less risk?
I talked with Ali a lot about MEDDIC, and now he’s an expert on all this stuff.
Turner Novak:
MEDDIC? I’ve never heard of MEDDIC before.
Ron Gabrisko:
MEDDIC is just a process for how you govern sales.
Turner Novak:
Oh, really? What is it?
Ron Gabrisko:
Each letter stands for a different thing. It’s like metrics: what metrics are you using to justify the deal, and how do they measure their business? What metrics are you actually changing with your solution? E is exec sponsor. D is decision process. Each letter means something in the process.
It’s a pretty well-known enterprise sales motion. There’s also a thing called Command of the Message, which is similar. Those are two frameworks I’d recommend for anybody trying to learn or build enterprise sales. They’re probably the two most common ones that I use and that we use. Then you branch off of that. It’s got to be custom for your company and your product.
Turner Novak:
Are there any mistakes you see people make when they’re first getting into this? Do they try to ramp up too quick, or what’s usually the biggest mistake?
Ron Gabrisko:
Companies or salespeople, or both?
Turner Novak:
Maybe it’s both. Maybe it’s the same thing. What’s usually the biggest mistake people make?
Ron Gabrisko:
The biggest mistake I see in salespeople is they start pitching before they understand anything. Like, I’m going to pitch you my product, but I don’t really understand you or your challenges. So first, get to know somebody, establish rapport, and ask questions, so you know what might be important to them.
From a company perspective, companies are growing faster now than ever. AI is a huge unlock for growing companies. I’m a big advocate that you need salespeople to grow your company, but the PLG motion has been great for companies like Anthropic. They’re doing great with PLG.
There are four stages of a company. There’s zero to 10 or 20 million, where you’re just finding product market fit. From 20 to 100 million, you’re building a playbook. It’s got to be a repeatable system with repeatable trends. From 100 million to a billion, you’re expanding internationally and into partners. And at multiple billions, it’s a lot about having the right leaders, culture, systems, and processes, and continuing that fast innovation. Each stage requires different things to grow.
Based on where a company is in its market, when I advise them, usually you need salespeople to push your product, because people don’t know about your product or know how to use it. And even when PLG takes off, you want salespeople to go talk to your enterprise customers. Enterprise customers are usually people who are buying stuff, and if people are buying stuff, you want people to sell stuff.
If it’s intuitive how to use something, then let it rip. Developers know how to use these coding tools, so awesome, they pick their favorite and go. You need salespeople to get through security reviews and contracts. But most products need salespeople to teach customers what the product does and how to get value out of it. So there are ways to ramp that up thoughtfully, depending on which market you’re going after.
Turner Novak:
I know you guys made a pretty strong bet that you were going to hire technical people to sell the product. Is that generally the rule, that the more technical the product, the more technical the salespeople have to be? You probably have to help the customers understand it, and you need the language to understand the problems.
Ron Gabrisko:
A thousand percent. My salespeople can demo our product themselves. We have an amazing pre-sales field engineering team, but everyone in my organization should be able to demo. They should know the product, because if you have technical buyers, you need technical sellers. Technical buyers don’t like non-technical sellers.
Turner Novak:
Is it because they feel like the seller doesn’t even understand what they’re talking about, so they can’t stand the person?
Ron Gabrisko:
If you don’t understand it, that’s even worse. But if you don’t bring anything to the table, why am I going to spend my time? My time’s precious. I’m only going to spend time with you if I’m going to learn something, or you’re going to help me solve a problem. So if I’m a technical buyer, you need to bring something that helps me solve my problem or get better or learn something. If you don’t, I’m not going to waste my time.
We have a very technical product. With Genie, it’s a lot simpler now, and we’re starting to sell to business users. But early days it was a pretty technical product, with technical buyers, data engineers, data scientists, data people. They’re highly technical. I need technical sellers to have credibility with them. That’s how I’d decide the profile.
As you scale an organization, you definitely need to know the profile of the seller that’s going to make the company successful. It’s not the same every time, but you need to build that profile and pattern for your company so you can scale it.
Turner Novak:
When you say profile, how do I understand what kind of profile I’d want?
Ron Gabrisko:
What are the main criteria I’m looking for in a seller for them to be successful here? You mentioned one, they need to be technical. I look for people who are technical, and who have experience in the data and AI market, from certain companies.
I also test for grit, perseverance, and hard work. I always say hard work overcomes talent every single day. We look for folks who understand how to sell consumption versus committed contracts, which usually means having some cloud experience. I like people who have some startup experience. Can they build? Do they know how to do things without a big machine behind them?
I also avoid people who have a bunch of short stints. Somebody who’s had two or three companies in less than two years, they don’t know what tough looks like yet. Every company hits tough times at some point, so you want people who have seen that and are going to dig in and push through.
I back-channel pretty hard. I look at every profile we hire, and I’m probably one degree of separation from everybody in Silicon Valley, so I can back-channel just about anybody. I’m looking for the best of the best, because I think this is a once-in-a-generation company.
Turner Novak:
When you talk about grit, determination, and hard work, how do you gauge that before someone’s actually done the job? They could say, oh, I work so hard. They could set their emails to send at 1:00 AM so it looks like they’re working. How would you actually gauge whether I’m a hard worker, whether I’m going to fight through it?
Ron Gabrisko:
It’s a little subjective. Obviously I’m looking for folks who haven’t had a bunch of short stints. But I also like to learn about the person and the character, like how they grew up, and what kinds of challenges they’ve had to work through.
My dad was a construction worker, my mom was a teacher. I grew up from nothing. So I want to hear somebody’s background and understand them as a person. We also give them assignments. We make them build a business plan. We make them pitch, like, give me your AI vision pitch for a CEO. We’ll give them the deck, but I want to see them pitch it. And we make them do a Genie demo for their vertical. We make them do some work to show they really want it.
I think if you get a job at Databricks, it’s a lottery ticket. And obviously, if I know people who have known them, I’ll ask, is this person going to be that caliber of seller? That’s why I look at startup experience too, because people in startups work crazy hard and have a lot of grit. We certainly did in the early days creating this company.
Turner Novak:
How do you do that kind of reference check? I might be good friends with someone, and you text me about them, and I say, oh, he’s awesome, you should hire him. What do you look for when you’re trying to get through the BS and understand what’s actually going on?
Ron Gabrisko:
Usually I’m checking with somebody I know and have a relationship with, so they’re not going to BS me, because they value mine.
Turner Novak:
So your relationship is above whoever the reference is. They value that with you.
Ron Gabrisko:
At least it’s, hey, if I tell Ron this person’s amazing, and then they’re not amazing, that’s not good. I’ve got a lot of one-degree-of-separation relationships with sales leaders, so they’ll be pretty honest, because a lot of the time they’re asking me too. They want an honest answer from me.
I’m not looking for a person who has nothing wrong with them. I’m trying to understand their strengths and the areas they need to work on. And I’ll ask, is this the top 1%, 5%, or 10% of people you’ve worked with? I’m trying to calibrate where they are on the bar. Some people I text will say, no, pass. So then you know. I don’t just check one, I check at least two or three, depending on the position. They’re pretty honest, because it’s a small circle.
Turner Novak:
You kind of have to know who you’re getting a reference from. Some of the founders I backed sold their company and did really well, and the references were some coworkers who said, I hate that guy, he’s so mean, he’s always bossing us around. He’d sold his first company to a big tech company, and he didn’t like the hierarchy and the meetings about meetings. He was just shipping stuff, and people in the company didn’t like it because he was trying to get stuff done. So some references said, this guy sucks, I don’t like working with him. You’re kind of calibrating the reference. In this case that’s actually good, because he was the CTO of a new company.
Ron Gabrisko:
You want to know what you’re getting into. All the feedback is good, but you’ve got to know the source.
Turner Novak:
Back when you were first building the sales team, it was pretty technical, and you hired a lot of sales reps. The number I saw was that you hired 40 reps in the first quarter. Is that true? That’s kind of insane.
Ron Gabrisko:
It’s a funny story. I would literally say, okay, all day Wednesday and Thursday, all I’m doing is interviews. They booked me, and I said, you’ve got to give me a break to go to the bathroom at least.
Turner Novak:
Okay, that’s good. At least you got that.
Ron Gabrisko:
Exactly. It was fun. A lot of those folks were from my network, people I trusted. My early thesis was, Spark is everywhere. My first task is to understand what they’re willing to pay for, and who I can sell it to. What’s the profile of the ideal customer? So you hire a bunch of people you trust to go talk to all those open source users and get that information and find the trends.
In today’s environment, with a lot of funding getting thrown around, that’s the first step. If you’re trying to sell a product, you need to hire a solid sales leader, or at least some salespeople you trust. Especially with an open source project, go talk to a bunch of customers who use the open source and understand what they’ll pay for. Am I selling to enterprise? Am I selling to startups? That’s job one.
We had just raised additional funding, so I did a coverage model to cover all the segments and find out which customers were more likely to buy, and what they wanted to buy. We moved pretty fast that first year. I think we went from less than a million to, I forget, $13 or $15 million, then we went to $50 to $100 to $250 million. Obviously now we’re $6.9 billion plus and bigger. But early days, those were known quantities for me, and my first task was to find out what people would pay for and which segments we could sell to.
Turner Novak:
So it wasn’t that you had no one using the product and hired a bunch of people to go get customers. You had millions of people already using it. So you were almost hiring for R&D and customer discovery, to go figure out what people were using it for, and then how you were going to make a business around it.
Ron Gabrisko:
Yeah. Initially we thought customers would just come to us and ask the questions. Before I got here, that’s what they were trying to do.
Turner Novak:
It wasn’t working?
Ron Gabrisko:
It wasn’t working, no.
Turner Novak:
What was the big blocker?
Ron Gabrisko:
Customers in open source don’t really want to buy anything. You have to ask them what they’re willing to pay for. It’s a little different. I’ll ask support questions, and once I get my question answered, I’m good to go. So salespeople open doors. They get in front of people. PLG is more about people coming to you. Sales is more about me coming to you. It’s the opposite. So sales will help open doors and open markets faster.
That said, if you have a great PLG motion, like ours was around open source, we had tons of open source folks coming in the door, and we needed people to go sell them the thing we actually monetize. So I’d recommend that for any open source project that has traction. Now, if you have a product and nobody’s used it, you need a couple of salespeople to go get your beta customers so you can also develop the product. That’s the zero-to-twenty-million, product-market-fit stage. You’re going to need a sales team to do that too.
Turner Novak:
I know you leveraged a16z for a big initial wave of customer introductions. How did you do that? A lot of people say, oh, use your investors to get customers, like it sounds super easy. How do you actually do that successfully?
Ron Gabrisko:
a16z is a little special. I think it’s the best VC on the planet, by the way. They build these startup teams that are like pieces of your company. They have sales, marketing, recruiting. And in this case, they have relationships with the CIOs of some of the biggest enterprises on the planet. They’ve built that for their portfolio.
They would reach out to a company like Apple or Cap One and invite the CIO and their staff to come in for a Silicon Valley day, and they’d show them ten different portfolio companies. We’d have thirty minutes to come in and do a demo. That way, those CIOs and their staff would see ten amazing startups in one trip. And they’d do it at a16z’s office, roll out the red carpet. It’s pretty cool, because you get to meet Ben Horowitz and Marc Andreessen. So they had a pretty good draw.
The mistake startups make, and these programs exist at some investors, and some run them better than others, is don’t just send any old salesperson. I did most of those early days myself. One, I get to pitch my product to the CIO of a big company, which is a big opportunity. And most of the time, the CIOs are so thankful you took the time to do it that they’ll give you an opportunity, a POC, or a small land. Now you’re in the door, and you have a big logo on a land.
So as a founder, go pitch those meetings yourself, the CEO, CRO, head of sales, make sure you take full advantage of them. Just saying, hey, make some intros, can you intro me to XYZ, is less valuable. I’d rather target a smaller number of companies where I can actually get a chance to talk to somebody and add some value. a16z, great program. We landed a bunch of our big customers that way.
Turner Novak:
It sounds like an in-person event where you get face time is the most efficient or best way to do it, versus just, can you forward an email.
Ron Gabrisko:
Exactly. If you want to ask your investor, I’m sure many investors do these, where they have a customer day and bring in ten portfolio companies to pitch to a panel of customers, or maybe one customer. This particular one was more each customer. That way you get face time and develop some relationships. What you’re trying to come out of there with is at least one follow-up, one relationship, one POC. So that’s what I’d ask for from an investor, versus just, hey, can you introduce me over the phone or on email.
Turner Novak:
Do you get invited to a lot of those things now? Are you guys considered a big customer now?
Ron Gabrisko:
Now I do a bunch of the reverse. A lot of investors say, hey, I have my portfolio coming in, can you do a fireside chat? Because a lot of these startups are technical, mostly technical founders. It might be their first company, it might not be, but it’s usually a technical-led founding team, and they’re trying to figure out, how do I build that sales go-to-market engine? It’s the number one thing I hear.
Most of the questions I get are similar to what we’re talking about here. How do I build that go-to-market team? How do I find that first sales leader? How do I land those first customers? How do I figure out pricing? How do I build the first comp plan? So a lot of my investors ask us to come in and talk with their portfolio CEOs about those topics.
Turner Novak:
When do you decide yes or no on those things? If I’m trying to do something like that, do I need an enticing pitch, like, it’s a cool venue, or a nice dinner? How do you decide what’s worth your time?
Ron Gabrisko:
Me personally...
Turner Novak:
I guess I’m trying to reverse this. If I was trying to set some of this up, what should I be offering to both sides of the marketplace to make it worth everyone’s time?
Ron Gabrisko:
Fair enough. Certainly as a portfolio company, I want access to customer executives, because I want to try to sell to them.
Turner Novak:
And you don’t want to feel like you’re getting sold to as the customer executive, maybe.
Ron Gabrisko:
Correct. On the customer side, for these investors it was, hey, come see the best of Silicon Valley. They’d ask, what do you want to see? Do you want to talk about AI? Security? They have a portfolio of different categories, so they’d cater the agenda and the portfolio companies, and let the customers pick, like, here are the companies I’m interested in.
I just hosted one big bank and one big healthcare company on the East Coast, brought their whole team, because they’re going to come to Silicon Valley and want to see Databricks and OpenAI and Anthropic and NVIDIA, if they’re in town, or the big hyperscalers. If your VC is a Silicon Valley VC, they should be able to help host them and get you a sit-down. So I’d try to host them in San Francisco that way. That’s what a16z’s model was, and it worked really well. It helped seed a lot of these portfolio companies into big customers, which is huge.
Turner Novak:
When you say that’s what the model was, do they not do it anymore, or is it different now?
Ron Gabrisko:
We’re kind of beyond that size. They’ve moved on to the smaller portfolio companies. Obviously we can get a lot of our own meetings now with CEOs and CIOs. We’re at multiple billions. But for early and mid-stage companies, that’s huge. I’ll still reach out to investors once in a while and say, hey, do you know so-and-so, can we get a meeting, if I’m having a tough time getting to somebody. So they can be super helpful. It’s definitely something you want to look at when you’re raising money.
Turner Novak:
When you talked about figuring out pricing with these early customers, should I be willing to do a pilot, or give a discount, to land some of the initial customers and get things going? How do you think about navigating that? It always comes up as, how big is the contract, what does it consist of, the timelines, all that.
Ron Gabrisko:
Most of our POCs and pilots, especially early days, were all free.
Turner Novak:
Oh, really?
Ron Gabrisko:
Yeah. We’re not trying to make money off the pilots or the POCs. We’re trying to prove value for our team and our product. Now, you probably want a couple of things. One, you don’t want it going on for a year, so set expectations around what you’re trying to prove. What are your success criteria? Two, what’s the time period? We’re going to do this over the next thirty days. And three, you want at least some executive sponsorship, so it’s not just a rogue developer saying, hey, help me do my project, and then you have no chance of selling it.
You need to prove value before you have an opportunity to ask for money and a purchase. That’s what a POC or pilot is really about. If you’re a bigger company with references, a brand, and a track record, you can use those, and you may not necessarily need a POC. But even then, I’d fund the POC or pilot if I know I’m going to get a bigger contract at the end. So I wouldn’t make the pricing of the POC the gating factor. Just make sure that once you’re successful, you have a good chance of actually getting a contract. That’s the more important point.
Turner Novak:
Because that first thing you land is probably not the final, right? If you do a good job, they’re going to spend more money, whether the initial pilot is paid or not.
Ron Gabrisko:
It’s usually a small piece. A small set of users, a small use case, one department. You’re trying to prove value and get your first land. It’s the land-and-expand model. Land, prove value, build a champion, expand, get all the rest of the use cases in that department, then start expanding to other departments until you’re across the entire enterprise.
Turner Novak:
As you started to climb the ladder, and people started to pay you, and Databricks really took off, was there anything surprising or unintuitive, maybe something you got wrong initially or changed, that you wouldn’t have expected as things scaled up?
Ron Gabrisko:
That’s a great question. There are like ten answers to that.
Turner Novak:
Okay, I want all of them. That’s actually a question from one of my portfolio company founders. She said, I want to know this.
Ron Gabrisko:
One thing I noticed is that when we started selling Databricks, we were doing $15K, $18K deals.
Turner Novak:
This is annual? That’s pretty small. People say, especially if you’re doing enterprise sales, that’s bad, or it could be better.
Ron Gabrisko:
Exactly. I said, new rule, no deals less than my monthly Uber bill. When I would talk to the customers, we were adding tons of value, but a lot of the use cases, because we’re usage-based, weren’t driving a lot of usage.
So we started to say, how do we capture some of that value? There are companies out there like Palantir that do value-based pricing. We weren’t doing that. We were doing straight cloud-based pricing. So we added a platform fee, then user fees, all these things to try to make our price points higher to match value.
Because your proper pricing model is, price equals value. If your price is over the value, no one’s going to buy it. If the price is under your value, you’re leaving money on the table. So basically, that’s what you’re looking for.
Turner Novak:
Is there a number? Is it like a 10x? Do you need to add 10x more value than you charge, or how do you think about it?
Ron Gabrisko:
Certainly there’s a payback. But some of this might be loose. If it’s, I’m going to grow your revenue by a billion dollars, I can’t charge you a percentage of that. So a lot of it is, how do you compare with the alternative, which is the competition or building it yourself? How do you compare with the lowest-cost alternative?
Once you get into it, pricing is complicated, but find out the base unit of value. For us, the clouds were charging on compute and usage, so that was the base value unit. Then it was relative to a cloud service, or building it yourself. We put all these features on to try to raise pricing early days.
One thing that wasn’t intuitive: once we started charging for users, people would restrict who used the product.
Turner Novak:
Oh, interesting. Which then also drives down usage.
Ron Gabrisko:
Drives down usage. Exactly. So I said, okay, let’s get rid of user pricing. All the users are free. And all of a sudden usage took off, because now everybody can use it and get value out of it.
A lot of people confuse users and usage. One of the best things we did early days was tie ourselves to consumption and usage, because data is growing, queries are growing, the number of people who want to ask queries of the data is growing, and the number of agents that want to ask queries is growing. All those things are tied to usage. Restricting yourself with user-based pricing, I think user-based pricing is a thing of the past, honestly.
Turner Novak:
Really? Is it just gone today? It just doesn’t make sense to do that anymore?
Ron Gabrisko:
The companies doing user-based pricing are under siege, because people aren’t growing employees. They might be growing agents, so maybe you charge for agents. But I think everyone is, or will, move to usage-based pricing. Even the coding tools were user-based early days. I remember talking to some of the early folks at Cursor, and I said, you guys need to go usage-based. Now that market has blown up with usage-based pricing.
So you’ve got to pick the base unit and test it out. I don’t think user-based pricing is a good idea. It limits the amount of value you can capture in most software. I can’t say that equivocally for everything, but attach yourself to something that represents value and is growing over time. It’s been a great business model for us, and for the clouds and all the frontier labs. They’re selling tokens, but it’s the same thing. It’s usage.
Turner Novak:
Because the revenue upside is essentially uncapped. If you keep creating more value, you can keep adding more revenue.
Ron Gabrisko:
Every day, there’s more data in my customer’s systems. There are more people asking queries, doing analysis, doing predictions, and more agents doing the same thing. So then you say, okay, what are the other pieces of TAM I can go after, internationally, different markets, different verticals? That’s how you start to compound exponentially, all those different markets together.
Turner Novak:
So that was one of the ten. Are there nine other unintuitive things? Anything else that really stands out, like, man, I wish I knew this back then?
Ron Gabrisko:
First of all, we talked about it a little, but just going out to the market and trying to understand what people will pay for and what their challenges are. You’d be surprised what customers will tell you when you just ask, hey, we’re building this company, what adds the most value, what would you pay for? A lot of people want to help.
Then, once you’re going into enterprise, a lot of companies will say, I’m just going to sell to the people who want to buy right now. Those people might be startups, AI companies, or tech companies, early adopters of technology. But if you want a super valuable company, trillions of dollars, you need to sell to enterprises. You need to sell to banks, healthcare companies, retailers, and CPG companies.
So one thing early-stage companies don’t think about is, what are those requirements around security and compliance? It’s a good idea to understand those upfront, because they’ll be big gating factors in how you grow. If you don’t understand how they affect your product upfront, it could slow you down quite a bit to add them later.
If you want to sell to the federal government, you need cleared personnel. You might need to quarantine those developers. There are all kinds of extra requirements. You should think about those things upfront. We missed some of that stuff, but it worked out.
Turner Novak:
Sounds like it worked out for you at the end of the day.
Ron Gabrisko:
It did. We made it so far.
Turner Novak:
Has anything changed about the go-to-market philosophy, strategy, or structure, how you sell, as the products became more AI-native and generative? You’ve always been consumption-based, which is one thing I hear a lot about with AI-native companies, that you’re selling usage versus seats. When you think about the move from on-prem to cloud, there was a change in what got sold. The incumbents got disrupted because they couldn’t just sell seats and renew the license. Then the shift from cloud to usage-based AI, again, they can’t quite sell it, because they’re used to selling a seat that no one uses, and now you actually have to use the product, and they can’t make as much revenue, and it messes up the business. How did that transition go for you? Or was it not even necessary?
Ron Gabrisko:
We were always cloud, always consumption. Some of the gating factors early days were, who’s in the cloud? I’d tell our sellers, if they’re not in the cloud yet, don’t waste a bunch of time, because AWS, Microsoft, or GCP has to convince them to get in the cloud before we can sell them anything.
Each stage has been a development for the go-to-market team. Early on, it’s about product-market fit and building a playbook. But then, how do you expand internationally? How do you expand with channels? We didn’t have partners early days. Now we do. So building out your partner channel, how do you build out EMEA? How do you build out APJ? Those are all developments.
And then, how do you go multi-product? Most companies start with one product. Now we have a pretty massive portfolio, so as you expand your portfolio and try to expand into additional areas of opportunity, that creates new muscles. How do you build out specialist organizations for each new product? And do all of that in a world where I’m asking, how do I AI-enable my entire sales team so they’re using it every day to be smarter and more productive?
All those things are changing constantly. It used to be you’d do an annual planning cycle. We were doubling or tripling every year, so I’d do a six-month planning cycle. We’re still on that, splitting territories every six months to make sure you have coverage on customers. Every piece of building and operating this business at this scale is new. There wasn’t a playbook for it. So it’s been fun.
Turner Novak:
Was there anything that messed up initially, or the biggest unlock, in how you expand internationally, and how you incorporate new products? Anything you wish you’d done differently, or something you figured out where you said, oh, this really unlocked it once we tweaked it?
Ron Gabrisko:
International is one area where you have to be careful about expanding too fast. You can do a lot of your initial sales remotely. You’re not going to build a multi-billion-dollar business internationally by selling from the Americas, but you can get some of your first lands. Because if you hire the wrong leader or the wrong strategy in EMEA or APJ, you’re on a 10 or 20-hour plane flight, and a lot of hours, to try to fix that.
So you need to find the right leader for each market. In EMEA, Germany is different from France, different from London, different from Southern Europe. Same with Asia, Japan, and India. Each is a different market, different language, different culture, so you need new leaders for each one. You need your playbook pretty set in the Americas before you go big internationally. Not saying you need to be at a hundred million, but somewhere between $20 and $100 million, you need a pretty good idea of what your sales playbook looks like, who you’re selling to, and what the requirements are, because you’re going to translate that into those other markets.
Having the right leader is part of it. But early days, I took some of my key talent from the Americas and said, hey, I’ll pay for you to live in London for a year and teach the new team. We try to do all this with enablement, but there’s always a lot of tribal knowledge. Same with APJ. Seed it with some of the people who have been successful and been here a while, so they understand it. That’ll help accelerate the growth.
Some companies make the mistake of going international too big, too fast, and then you spend a lot of money, and sometimes it hurts your brand. If you have a bunch of people who don’t know how to sell your stuff, or aren’t making customers successful, it’s harder to address later. So my advice is, be aggressive, but be thoughtful when you expand internationally.
Turner Novak:
So it sounds like, bring some of the local people who have been there and really know the product, the edge cases and pain points, but also hire somebody who knows Japan, or knows India, who says, this is how they do it here, don’t forget you have to do this one specific thing.
Ron Gabrisko:
You want the local leader, because they know which great salespeople to hire, and they know the customers. And you want some experienced knowledge alongside them. The local leader will probably learn it, it’ll just take them a year. When you put the experienced person with them, they go that much faster. So if you can do it, I’d recommend it.
Turner Novak:
Talking about actual enterprise AI adoption, what are you seeing today? What kinds of challenges? Where are people having the most success? You probably have as close to a front-row seat as you can get. What’s actually going on right now?
Ron Gabrisko:
I’ve literally talked to thousands of customers. We did publish numbers, like $1.7 billion just in AI.
Turner Novak:
In revenue, $1.7 billion in AI revenue?
Ron Gabrisko:
Yeah. It’s been growing super fast. It’s every industry. I call it the four C’s of what’s important for a lot of these companies.
The first, we talked a little about, is context. How do I attach AI to my data? It needs the context of my data to be smart about the decisions and predictions I want to make for my business.
The second is control. I need it governed. I can’t have everyone having access to all the data, especially in the world of models and agents. I need to know how these agents are going to use the data, and make sure they’re not disclosing it. There’s a lot of compliance in regulated industries, so control is incredibly important.
Then choice. Right now we’re seeing a lot around model choice, but cloud choice is also important. We’re open source, so you can plug anything into the Databricks platform, and we serve all the models. Frontier models are really good for a lot of the complex tasks, and open source models will be used for a lot of the other tasks, and that market’s going to grow super fast. Choice is really important, because if you get locked into one, you’re going to end up spending a lot of money.
And that’s the last thing, cost. Costs have been through the roof on a lot of this stuff, so how do you govern those costs? We do all of that in a product we call Unity AI Gateway, to give it a plug. A lot of these CIOs and CEOs are blowing through their budgets super fast. So how do you put on the right cost controls and guardrails?
That’s what I’m seeing in the market. But the use cases are phenomenal. I’m seeing new drugs discovered and accelerated, all kinds of financial services use cases around automating loan origination and fraud, retailers maximizing revenue with campaigns and promotions, how they stack their shelves, how they do distribution, how they fulfill inventory faster. It’s every industry. The market’s growing super fast. The biggest part is, the models are super smart, but how do you get the context of your enterprise data, and connect those things in the best, most efficient, most governed way?
Turner Novak:
What seems to be the biggest challenge these guys are facing? Is it that it’s expensive? Is it that they don’t know what to do?
Ron Gabrisko:
It’s a lot of the expertise. Everybody’s got an FDE model. We do too. I think ours are the best on the planet, but they need some help. They either need help from us, or from the frontier labs or the consulting companies, to wire all this stuff together. It’s not simple. We’re trying to make it as simple as possible with a bunch of our new tech, but it still requires some expertise.
The biggest challenge is that everybody recognizes the data is key to how you do these enterprise use cases, but in a lot of cases, the data is not in the right place yet. It’s all over the place. It’s in legacy systems, old formats, or proprietary formats. So getting your data in a good place to attach AI is a tough, complicated problem, and I think Databricks is the best on the planet to do that.
Turner Novak:
Is there any irony to it? AI’s doing all this stuff, and we can’t just say, hey AI, figure out how to do this for me. It’s kind of funny that it does all this stuff, and people are still struggling to use it correctly.
Ron Gabrisko:
It’s getting there. Genie Code can now start building data engineering pipelines for you. It does all the ontologies, so it’ll go find the datasets and label them. And you want to do all that in a governed fashion, which we use Unity Catalog for. AI is getting smarter about how it helps with data problems. That’s the main use case for us for Genie Code, which is part of Genie. We have a data engineering Genie Code, a data science Genie Code. We’ll go develop that stuff and automate it. So it’s getting there.
Turner Novak:
When you say your FDEs, your forward deployed engineers, are some of the best on the planet, what makes them so good? Do you structure them as part of the sales team? Are they using Databricks as the product, which makes them better? What makes them so good?
Ron Gabrisko:
You’ve got to start with understanding what the customer is trying to accomplish, which is similar to any consulting project. But then, being a thought leader in how to solve these problems, which technologies, the best and most efficient way, and the downstream effects.
A lot of companies use FDEs. Palantir uses that model, lots of people use it. One thing you don’t think through is the ongoing upgrading and maintenance of what you build. On Databricks, we’re a data and AI platform, so we’re thinking through how these solutions develop and evolve over years and years. They can’t have a huge army trying to upgrade and maintain them. That’s probably a great revenue model for some, but for us, we’re thinking through how this thing evolves and grows over time, and we let the actual team at the customer maintain it. That’s why you need a scalable, unified data platform underneath all of this.
There are lots of talented technical FDEs who can solve these problems. But how you solve them in the right way, where it’s scalable, cost-efficient, and evolves and grows over time without having to rewire everything, I think those are the keys, and that’s what makes our team the best.
Turner Novak:
Do they report to engineering, or to sales? I’ve always wondered what the right way to structure that is, because they’re kind of both.
Ron Gabrisko:
We have a group we call field engineering. That’s all of our pre-sales, the solution architects doing all our pre-sales architecture work. Our FDEs sit inside that organization, so they go deep on the product. Those folks are engineers. They can code. They can build products, build out the pilots, build out the environments. We call them field engineering. That reports up into go-to-market, but it’s run by one of our co-founders, Arsalan. He’s also a PhD guy, so they’re crazy smart. They’re awesome.
Turner Novak:
I have to ask you this, because anyone listening at this point is like, you’ve got to ask him about this. Databricks is a $188 billion company. When do you guys go public? You just raised a couple billion more dollars, so you could have done it. When does that actually happen? Because there are some companies that want to stay private forever, it seems. What’s the view inside Databricks?
Ron Gabrisko:
It’s not a matter of if, it’s a matter of when. We run this company like a public company. We report our financials, and every quarter we have a board meeting where we go through our audit committee. We run it like a public company.
We’re not here to... We’re here to build a trillion-dollar company. I’d say we’re going public six months at a time. It’s not a matter of if, it’s just when. And we’re not in a rush. We’re going to build this and make it a trillion-dollar company. That’s the journey.
Turner Novak:
That’s a big goal, but I guess you guys are getting closer and closer, to where it’s like, you’re actually pretty close at this point.
Ron Gabrisko:
I’m going to start saying we’re going to be multi-trillions.
Turner Novak:
Yeah, you’re going to have to adjust, for people who are like, come on, only a 5x from here? You’ve got to shoot higher than that.
Ron Gabrisko:
Exactly.
Turner Novak:
Do you have a favorite business CEO or founder? You can’t say Databricks, or anyone at Databricks. Or a historical figure you get a lot of inspiration from.
Ron Gabrisko:
I love a lot of sports heroes. But I was going to say LL Cool J, actually. I started a business with LL Cool J way back when.
Turner Novak:
Really? I did not know that.
Ron Gabrisko:
An interesting company. They did this virtual recording studio, where a kid could record a song from, it was kind of pre-Skype back then, a kid could record from LA and New York together. The guy’s reinvented himself many times, in many different industries, an icon over many decades. And he’s a good friend, so I thought I’d throw him out there.
Turner Novak:
Interesting. What is he up to today? I don’t follow him that closely.
Ron Gabrisko:
He did his show, NCIS: Los Angeles. He was doing a bunch of TV stuff.
Turner Novak:
He did that for like 14 years?
Ron Gabrisko:
Yeah, exactly.
Turner Novak:
That’s great. Do you have a favorite athlete, then? You said there were some athletes.
Ron Gabrisko:
Michael Jordan, for sure, is my favorite. The GOAT of basketball, in my opinion. You watch his story. There’s a book called Relentless, by the guy who trained Michael Jordan and Kobe Bryant, and a lot of those elite basketball players. It’s a good book. So Michael Jordan, maybe. And Walter Payton. Love Walter Payton. I’m a Bears fan, grew up in Chicago.
Turner Novak:
I don’t actually know the Walter Payton story that well. What was his journey to the NFL, and what made him so good?
Ron Gabrisko:
Walter Payton was on the Bears, who were always one of the worst teams. They never had a quarterback, so they would just hand him the ball every time, and he would still break records running. To train, he would run up hills. He ended up dying of cancer, early. But he always won the humanitarian award. There’s an award now, I think the Walter Payton Award, for the best humanitarian in the NFL. He always gave back to kids and people who were struggling. That made him a hero of mine.
Turner Novak:
It’s good to remember where he came from, and to help people along the way.
Ron Gabrisko:
Exactly.
Turner Novak:
Ron, thanks for coming on the show. This was a lot of fun.
Ron Gabrisko:
Super fun. Thanks for having me.
Turner Novak:
Do you have any crazy stories? Anything just crazy that’s happened in your life that no one would believe?
Ron Gabrisko:
The LL Cool J story is pretty good. A lot of people don’t know I started a company with that dude.
Turner Novak:
What happened with it? Did you get people using it? Did you sell it?
Ron Gabrisko:
We got hundreds of thousands of people using it. We licensed it to Sony, Microsoft, Dolby, and a bunch of that stuff, and I think it’s still out there. But we never updated the technology. The technology is way more advanced now. We did that back in like 2013, so it was a long time ago.
It was super fun. I’ve been on stage with him, singing. I’ve been to a couple of Grammys where he hosted. It was pretty fun. A different world than tech software, for sure.
Turner Novak:
Different world.
Ron Gabrisko:
Exactly. I talk a lot about my upbringing, as a Midwest kid out of Chicago. Dad’s a construction worker, mom’s a teacher. I think hard work and hustle. I always wanted to be a Major League Baseball player. I played in college, I was an All-American, Big Ten Medal of Honor, but I never made it.
I’ve always kept that hard work, grit, and hustle. It’s a big cultural thing for me with the sales team here. I talk about it a lot, because I want the next generation of sellers, business people, and entrepreneurs to realize that nothing comes easy. Even if it looks easy, nothing comes easy, because it’s always about who out-hustles. I always say, I never lost a game, the clock just ran out. If we kept going, I would’ve figured out a way to go further than the next person.
Turner Novak:
What do you think separates the people who can do that from the ones who can’t? Is it an internal drive? What is it?
Ron Gabrisko:
I think it’s an internal drive, and I don’t think it’s externally motivated. People ask me, why don’t you retire yet? And I say, because one, I’m having a ton of fun and I’m passionate about what we’re doing, but I’m not done. Until we make this one of the greatest companies on the planet, if not the greatest, I’m going to keep going. So it’s internally motivated. It’s a drive.
Again, from that Relentless book, it talks a lot about that. It talks about cleaners and closers, and the difference between the two.
Turner Novak:
What’s the difference? I’m looking up Relentless right now. I’ll throw a link in the description for people to check out the book.
Ron Gabrisko:
Well, I don’t know if it’s a great recommendation. I don’t know if it’s the most PC book. Maybe it is, I don’t know. Dara can decide. But it’s the difference... Did you ever watch The Last Dance with Michael Jordan? It’s about the three-peats, about the Bulls.
Turner Novak:
I did watch it. I’ve seen it, yeah.
Ron Gabrisko:
They talk about his will to win. He almost didn’t even need a coach, because he needed a coach just to make sure everybody else didn’t quit.
Turner Novak:
Oh, wow, okay.
Ron Gabrisko:
He was that hard on his teammates. He wanted to win that much. There’s one page in that book that kind of explains the whole book. It’s a lot of those kinds of things. Like, when everybody’s hitting the panic button, they all turn to you. It’s about being the best of the best.
Turner Novak:
So it’s about being that person when everyone hits the panic button. Who do they go to? You want to be that guy.
Ron Gabrisko:
You want to be that guy. You want to take the last shot, all those things. Which is me and Michael Jordan, right?
Turner Novak:
Well, cool. This has been a lot of fun. Thanks for coming on the show.
Ron Gabrisko:
It was fun. Great spending a couple of hours with you. Appreciate it. It was awesome. Can’t wait to see it, can’t wait to hear it.
Find transcripts of all other episodes here.
