AI x CIO: Naveen Zutshi on the Changing Role of the CIO

Hello, hi everyone. I'm Abhijit Mitra, CEO of Outreach. Outreach is the AI revenue workflow platform. And it's my extreme pleasure to have with me today, Naveen Zutshi, CIO of Databricks. Databricks, the data intelligence platform. So Naveen, would love to hear a little bit about yourself or our audience. Thank you, Abhijit. Great to connect with you today. My name is Naveen Zutshi. I've been with the company close to four years now. I used to be the CIO of Palo Alto Networks before this. At Databricks, I'm responsible for applications, infrastructure, building some of the AI workflows that we're gonna talk about today, as well as we look to automate in a rapidly growing company. For those who don't know about Databricks, we have, our mission is to democratize data and AI. And we do this by providing a platform that works on all the three major cloud providers. And essentially we have 15 plus thousand customers now, and we're growing at 4 billion run rate. Amazing. So with AI being a corporate band-aid, CIOs like yourself are actually in charge of most organizations AI strategy. And that's what's the genesis of this CIO AI series for our audience. That's why we're doing this interview with you Naveen. You know, as a CIO, you see a lot across the entire business because you touch pretty much every function in your company. What opportunity does AI bring to CIOs and IT teams? And what are some of the challenges in that? What's your view on that? Yeah, the hype has been massive. And at the end of the day, we are focused on outcomes. So how can AI drive incremental revenue to our business and how can it make our employees a lot more productive? from just summarization or classification or discovery of information to actually taking actions. And those actions could be as simple as, hey, I'm having a PTA request coming from an employee and I want to evaluate before I approve that PTA request and then automatically approve it based on my decision, based on the criteria that I check, or it could be more complex around, hey, I want to improve the way we do our sales outreach to our customers, yeah. And that's the context of where we have a joint shared interest actually, because for our viewers, Outreach is an AI revenue workflow platform. So we use different types of AI, most recently, agentic AI to orchestrate workflows, revenue workflows, which brings me to my next question, which is, how do CIOs prepare for this rapid introduction of AI and specifically AI agents into their revenue workflows? Yeah, I think there is a lot of enthusiasm around that. And the way we thought about is, at least for what we have seen our customers think about it as well, is look at the lead to cash workflow for an enterprise. We are an enterprise B2B, we are more of a consumption-based company. Many companies are more commit-based, but regardless, the problem statement is, hey, I have a lot of prospecting to do, and I want to be very efficient in that. I have a lot of leads to nurture, I want to be very efficient and effective use of my marketing dollars for my lead management. And then that goes into prospecting. And then once I have prospect identified, I want to be able to spend more time as a sales rep on customers rather than on administrative trivia. As an example, you talk to salespeople and this is supported by data as well, that 70 to 80% of the time the sales reps are spending is on administrative work, either entering information into systems or answering questions from their managers or just building reports or building proposals for their customers, rather than actually meeting customers and really solving the problems of the customer. And our thesis is we want to spend, we want our sales teams to be spending more time with customers rather than on the administrative trivia. So that was the thesis behind what we have been working on. And we can talk about a few things that we have built there or are in the process of building as well. Yes, yes, let's talk about that because that's very exciting for our audience and for me in particular. And I've seen some interviews you've done, you've done a lot actually there. So we'd really love to hear about how you're making it easier for and sort of saving that 70 to 80% of the administrative time for reps so that they can spend more time with their customers. In fact, we have a saying here, it's just like, let reps do what they do best and let AI do the rest. And what is that about what they want to do best? Is meeting customers, closing these building relationships, having that strategic discussion. So how are you doing that at Databricks? Yeah, I'll talk about three main areas of investment that we have made and levels of maturity. And so one is the most mature and then subsequent, and then subsequent, right? So the first one, we built what is known as a sales analytic hub. It is a combination of something we build with business where we use the data intelligence platform of Databricks to pull in all the key data for our sales organization from Salesforce, from Clari, from Outreach, from other tools as well. And it all feeds into this sales analytic hubs. And essentially our sales reps all the way to regional managers, to country leaders, to our CRO, uses the sales analytic hub to run their business. So it has really helped in managing our business a lot better rather than having to go to five different solutions. And so most of the leaders are spending a lot of time there. Second, for top of the funnel, which is essentially SDRs and BDRs, folks who are really focused on prospecting and really hunting, they were spending a lot of time on contact scraping, account creation, understanding where the contact is in terms of their journey. For example, they would do LinkedIn scraping, they will look at the websites, they will try to search for information about those contacts so that they could reach out to those contacts, then they spend a lot of time crafting the right emails. And we essentially built three agents that work together and they are the writer, the researcher and the creator. So those agents work together essentially to help you build these emails and in the context of the prospect. And what we saw is with those emails, the click through and response rate, response rate as an example is 6% compared to 4.5% for an average BDR. And we are seeing up to 20% productivity improvement for our SDRs and BDRs. So what used to happen is we would have a lot of leads that we just couldn't handle because of the demand. And our SDRs and BDRs are more productive and they're spending more time talking to customers rather than spending all the time scraping contacts. So and then the third example is one we are building on the data intelligence platform at Databricks is for middle of the funnel. So if you take the sales reps, we have 4,000 people in sales organization between field sales and technical field engineers. And they're spending a lot of time on creating EBC doc reps or creating three option proposals or creating quotes or updating contacts and accounts. They spend all this time on what I would say data that we can make them a lot smarter on. So we are building these sales agents, which essentially are part of this one simple full stack app that is built on Databricks, which enables you to do those activities with an agentic first mindset. And so these are something we are really excited about. We feel the third use case that I'm sharing about is probably the earliest, but will have the most impact to us because it has the most number of employees behind it and also will impact revenue in significant manner if we do it right. That's awesome. So again, for the viewers, Naveen is our Databricks as a partner of Outreach. And we are actually super excited to be part of the journey of some of the things that Naveen is talking about. One of the things I do want to ask you, Naveen, is that as you evaluate various technology and also the stuff that you're working on in-house, how do you figure out which use cases and where you should build in-house versus buy off-the-shelf technology from a standard packet software vendor? So how do you decide on that? Yeah, I think the answer to this question varies between buy the company. So one of the things maybe unusual about Databricks is we are more of an engineering-led company. And so our bias is towards building more. That doesn't mean we don't have SaaS apps, we do. But our thesis around SaaS has been also that we feel that they are disjointed and fragmented today in most enterprises. You have hundreds of SaaS apps, in some cases thousands of SaaS apps, and there is a lot of suitable armchair that happens for every persona going across all these. So now we are not going to replace all those SaaS apps. We may replace some. And what are we going to do instead? So what our thesis is we will build some outcome-based agents, which will simplify some of those personas. And then in other cases, we may buy where the technology is mature and the result set is very clear that we see the win. So for example, the example that you actually mentioned, where we are using Outreach with a combination of another solution, we found that in the flow of the use cases and sequences that the BDRs and SDRs used, having Outreach plus this other tool was really beneficial. And we didn't have to build it because one, we saw the efficacy of the results. The results were great during our POC step. And then technology providers like you were really helping us build with us, right? So our forward leaning on that. Outreach also includes data lake as part of its solution for all our customers. And that is actually running on data breaks. And the stuff that we talked about is what we do for getting all the structure and unstructured data into the outreach data cloud so that our AI agents can actually take action on that. So that's how we do it for our customers and for ourselves as well. So my last question for you would be, what advice do you have for other CIOs as they start their AI journey today? One, involve business deeply in from the start. I think this is one example where there's a lot of pull from the business. So whether that is business and IT hackathons together, do that. Second, make sure you have a team in IT that can actually do the work. So one of the things we discovered is we had to build a platform group that can actually deliver on the AI promise. So it doesn't have to be a large team, but it has to be strong software developers with the pension for building AI solutions and coupled with data scientists and product managers. Third, find ways to create a sandbox environment which you can use for building agents on the edge. Provide a solution for customers to democratize because your business teams are going to be after it. And if you don't do it, there will be shadow IT and they will do it themselves. So find ways to make them successful as well. And then fourth, do it all in the framework of legal governance and governance. Legal governance and security. Really make them partners of yours. I work with our chief legal officer and our security officer and our engineering team very, very closely. In fact, we cross-pollinate our teams. We have folks from R&D join IT to help with the agentic work. So do that as well. So that will just help you accelerate the journey. And then fifth, I think prioritizing on use cases. Really do the feasibility versus value exercise. What are the hard metrics that you can drive and how feasible is the solution? No rocket science here. Nothing I've said here in the five things I've mentioned are not known to CIOs. You just have to do them. Awesome. And with those words of wisdom, we will end our interview today. Thank you, Naveen, for spending the time with us and sharing with us your experience with AI and CIO of Databricks. Thank you so much. Thank you, Abhijit. Take care.

Hello, hi everyone. I'm Abhijit Mitra, CEO of Outreach. Outreach is the AI revenue workflow platform. And it's my extreme pleasure to have with me today, Naveen Zutshi, CIO of Databricks. Databricks, the data intelligence platform. So Naveen, would love to hear a little bit about yourself or our audience. Thank you, Abhijit. Great to connect with you today. My name is Naveen Zutshi. I've been with the company close to four years now. I used to be the CIO of Palo Alto Networks before this. At Databricks, I'm responsible for applications, infrastructure, building some of the AI workflows that we're gonna talk about today, as well as we look to automate in a rapidly growing company. For those who don't know about Databricks, we have, our mission is to democratize data and AI. And we do this by providing a platform that works on all the three major cloud providers. And essentially we have 15 plus thousand customers now, and we're growing at 4 billion run rate. Amazing. So with AI being a corporate band-aid, CIOs like yourself are actually in charge of most organizations AI strategy. And that's what's the genesis of this CIO AI series for our audience. That's why we're doing this interview with you Naveen. You know, as a CIO, you see a lot across the entire business because you touch pretty much every function in your company. What opportunity does AI bring to CIOs and IT teams? And what are some of the challenges in that? What's your view on that? Yeah, the hype has been massive. And at the end of the day, we are focused on outcomes. So how can AI drive incremental revenue to our business and how can it make our employees a lot more productive? from just summarization or classification or discovery of information to actually taking actions. And those actions could be as simple as, hey, I'm having a PTA request coming from an employee and I want to evaluate before I approve that PTA request and then automatically approve it based on my decision, based on the criteria that I check, or it could be more complex around, hey, I want to improve the way we do our sales outreach to our customers, yeah. And that's the context of where we have a joint shared interest actually, because for our viewers, Outreach is an AI revenue workflow platform. So we use different types of AI, most recently, agentic AI to orchestrate workflows, revenue workflows, which brings me to my next question, which is, how do CIOs prepare for this rapid introduction of AI and specifically AI agents into their revenue workflows? Yeah, I think there is a lot of enthusiasm around that. And the way we thought about is, at least for what we have seen our customers think about it as well, is look at the lead to cash workflow for an enterprise. We are an enterprise B2B, we are more of a consumption-based company. Many companies are more commit-based, but regardless, the problem statement is, hey, I have a lot of prospecting to do, and I want to be very efficient in that. I have a lot of leads to nurture, I want to be very efficient and effective use of my marketing dollars for my lead management. And then that goes into prospecting. And then once I have prospect identified, I want to be able to spend more time as a sales rep on customers rather than on administrative trivia. As an example, you talk to salespeople and this is supported by data as well, that 70 to 80% of the time the sales reps are spending is on administrative work, either entering information into systems or answering questions from their managers or just building reports or building proposals for their customers, rather than actually meeting customers and really solving the problems of the customer. And our thesis is we want to spend, we want our sales teams to be spending more time with customers rather than on the administrative trivia. So that was the thesis behind what we have been working on. And we can talk about a few things that we have built there or are in the process of building as well. Yes, yes, let's talk about that because that's very exciting for our audience and for me in particular. And I've seen some interviews you've done, you've done a lot actually there. So we'd really love to hear about how you're making it easier for and sort of saving that 70 to 80% of the administrative time for reps so that they can spend more time with their customers. In fact, we have a saying here, it's just like, let reps do what they do best and let AI do the rest. And what is that about what they want to do best? Is meeting customers, closing these building relationships, having that strategic discussion. So how are you doing that at Databricks? Yeah, I'll talk about three main areas of investment that we have made and levels of maturity. And so one is the most mature and then subsequent, and then subsequent, right? So the first one, we built what is known as a sales analytic hub. It is a combination of something we build with business where we use the data intelligence platform of Databricks to pull in all the key data for our sales organization from Salesforce, from Clari, from Outreach, from other tools as well. And it all feeds into this sales analytic hubs. And essentially our sales reps all the way to regional managers, to country leaders, to our CRO, uses the sales analytic hub to run their business. So it has really helped in managing our business a lot better rather than having to go to five different solutions. And so most of the leaders are spending a lot of time there. Second, for top of the funnel, which is essentially SDRs and BDRs, folks who are really focused on prospecting and really hunting, they were spending a lot of time on contact scraping, account creation, understanding where the contact is in terms of their journey. For example, they would do LinkedIn scraping, they will look at the websites, they will try to search for information about those contacts so that they could reach out to those contacts, then they spend a lot of time crafting the right emails. And we essentially built three agents that work together and they are the writer, the researcher and the creator. So those agents work together essentially to help you build these emails and in the context of the prospect. And what we saw is with those emails, the click through and response rate, response rate as an example is 6% compared to 4.5% for an average BDR. And we are seeing up to 20% productivity improvement for our SDRs and BDRs. So what used to happen is we would have a lot of leads that we just couldn't handle because of the demand. And our SDRs and BDRs are more productive and they're spending more time talking to customers rather than spending all the time scraping contacts. So and then the third example is one we are building on the data intelligence platform at Databricks is for middle of the funnel. So if you take the sales reps, we have 4,000 people in sales organization between field sales and technical field engineers. And they're spending a lot of time on creating EBC doc reps or creating three option proposals or creating quotes or updating contacts and accounts. They spend all this time on what I would say data that we can make them a lot smarter on. So we are building these sales agents, which essentially are part of this one simple full stack app that is built on Databricks, which enables you to do those activities with an agentic first mindset. And so these are something we are really excited about. We feel the third use case that I'm sharing about is probably the earliest, but will have the most impact to us because it has the most number of employees behind it and also will impact revenue in significant manner if we do it right. That's awesome. So again, for the viewers, Naveen is our Databricks as a partner of Outreach. And we are actually super excited to be part of the journey of some of the things that Naveen is talking about. One of the things I do want to ask you, Naveen, is that as you evaluate various technology and also the stuff that you're working on in-house, how do you figure out which use cases and where you should build in-house versus buy off-the-shelf technology from a standard packet software vendor? So how do you decide on that? Yeah, I think the answer to this question varies between buy the company. So one of the things maybe unusual about Databricks is we are more of an engineering-led company. And so our bias is towards building more. That doesn't mean we don't have SaaS apps, we do. But our thesis around SaaS has been also that we feel that they are disjointed and fragmented today in most enterprises. You have hundreds of SaaS apps, in some cases thousands of SaaS apps, and there is a lot of suitable armchair that happens for every persona going across all these. So now we are not going to replace all those SaaS apps. We may replace some. And what are we going to do instead? So what our thesis is we will build some outcome-based agents, which will simplify some of those personas. And then in other cases, we may buy where the technology is mature and the result set is very clear that we see the win. So for example, the example that you actually mentioned, where we are using Outreach with a combination of another solution, we found that in the flow of the use cases and sequences that the BDRs and SDRs used, having Outreach plus this other tool was really beneficial. And we didn't have to build it because one, we saw the efficacy of the results. The results were great during our POC step. And then technology providers like you were really helping us build with us, right? So our forward leaning on that. Outreach also includes data lake as part of its solution for all our customers. And that is actually running on data breaks. And the stuff that we talked about is what we do for getting all the structure and unstructured data into the outreach data cloud so that our AI agents can actually take action on that. So that's how we do it for our customers and for ourselves as well. So my last question for you would be, what advice do you have for other CIOs as they start their AI journey today? One, involve business deeply in from the start. I think this is one example where there's a lot of pull from the business. So whether that is business and IT hackathons together, do that. Second, make sure you have a team in IT that can actually do the work. So one of the things we discovered is we had to build a platform group that can actually deliver on the AI promise. So it doesn't have to be a large team, but it has to be strong software developers with the pension for building AI solutions and coupled with data scientists and product managers. Third, find ways to create a sandbox environment which you can use for building agents on the edge. Provide a solution for customers to democratize because your business teams are going to be after it. And if you don't do it, there will be shadow IT and they will do it themselves. So find ways to make them successful as well. And then fourth, do it all in the framework of legal governance and governance. Legal governance and security. Really make them partners of yours. I work with our chief legal officer and our security officer and our engineering team very, very closely. In fact, we cross-pollinate our teams. We have folks from R&D join IT to help with the agentic work. So do that as well. So that will just help you accelerate the journey. And then fifth, I think prioritizing on use cases. Really do the feasibility versus value exercise. What are the hard metrics that you can drive and how feasible is the solution? No rocket science here. Nothing I've said here in the five things I've mentioned are not known to CIOs. You just have to do them. Awesome. And with those words of wisdom, we will end our interview today. Thank you, Naveen, for spending the time with us and sharing with us your experience with AI and CIO of Databricks. Thank you so much. Thank you, Abhijit. Take care.