David Totten on this episode
David Totten is VP of Field Engineering at Vercel, a $9.3 billion platform powering much of the modern web for customers including OpenAI, Nike, Netflix, and Walmart. With nearly 20 years at Microsoft and a background in cloud infrastructure, Totten leads every technical touchpoint in Vercel's customer journey from pre-sales exploration through post-sale optimization and support.
Vercel treats go-to-market like a product, deploying AI agents across every stage of the customer lifecycle to eliminate manual work and surface real signals for faster decision-making. The shift from reactive GTM ops to proactive, AI-driven workflows has fundamentally changed how Vercel scales technical expertise without adding headcount. Rather than scaling engineering teams linearly, Vercel automates the knowledge capture, customer feedback synthesis, and churn prevention that typically require manual management.
The reframe here is simple but profound: if you're asking technical people to spend time in CRM systems or manual documentation, you're not using your resources optimally. Eve, Vercel's agent framework, emerged from customer demand to democratize agent building. A sentiment analysis agent replaced hours of CRM data entry per week. Usage telemetry across millions of deployments surfaces churn signals before customers realize they're at risk. AI Gateway helps customers optimize foundation model selection for cost and quality, unlocking hidden expansion revenue. The philosophy binding it all together is that post-sale is really just continuous expansion—every customer can use more of Vercel than they currently are.
The episode walks through how this works operationally: how Vercel feeds customer support tickets into a product roadmap, how field engineers move from reactive problem-solving to proactive recommendation engines, and what a 12-month journey from 80% reactive to fully proactive GTM looks like at scale.
What we cover in this episode
-
2:37
Role of VP Field Engineering Owns all technical touchpoints from pre-sales discovery through post-sale optimization and support.
-
5:04
From Manual to AI-Driven GTM Consolidated pre and post-sales with forward-deployed engineers and automated support agents to reduce friction.
-
11:55
Eve Agent Framework Agent framework that lets anyone build agents for internal and external workflows with governance, security, and observability.
-
16:52
Treating GTM Like a Product Customer feedback drives product roadmap. Build features from validated customer use cases, not internal assumptions.
-
31:34
Sentiment Analysis Agent Eliminates CRM Work Runs on sales calls and support tickets to auto-populate CRM, capture feedback, and surface product recommendations without human data entry.
-
40:18
Post-Sale Expansion and Churn Prevention Monitor usage telemetry, flag cost anomalies, recommend optimization patterns in product to prevent churn via architecture guidance.
-
42:31
AI Gateway for Cost Optimization Helps customers select optimal foundation models for cost and quality trade-offs, unlocking expansion revenue.
-
46:16
Usage Telemetry Surfaces Churn Early Millions of deployments generate signals that predict churn before it happens by tracking spend patterns and workload changes.
The lines worth sharing
Our job is to display the art of possible with customers on what can be done with for sale, and give an opportunity to take advantage of the latest tech trends.
David Totten · 1:12
We build 98% of the products off of a customer's use case. We are very customer driven, use case oriented.
David Totten · 9:11
30% of the applications being deployed on our platform are now agent-built. Six months ago, 2% were agent-built. Now it's over 50%.
David Totten · 16:50
We don't want bad revenue. We don't want customers spending a bunch of money with us for bad architecture. Some churn is by design, right?
David Totten · 47:56
Common questions from this episode
What is a VP of Field Engineering?
Owns all technical customer touchpoints from pre-sales through post-sale support. Responsible for demonstrating product capabilities, managing proofs of concept, deployment optimization, technical support, and gathering customer feedback for product development.
What is Eve and how does it work?
Eve is Vercel's agent framework that lets anyone build agents for internal or external workflows. It provides governance, security, observability, and automatic integration with databases and identity systems without manual provisioning.
How does Vercel use sentiment analysis to eliminate CRM data entry?
A sentiment analysis agent automatically captures and categorizes feedback from sales calls and support tickets, then updates the CRM via API without manual entry. Engineers focus on customer work while the agent handles data entry and flagging insights for leadership.
How does Vercel detect and prevent churn?
Usage telemetry across millions of deployments flags cost anomalies, spend rate changes, and workload pattern shifts. The system recommends optimization patterns and cheaper alternatives in product to address churn drivers early.
What is AI Gateway and why does it matter for expansion?
AI Gateway helps customers select the right foundation model for their use case based on cost and quality trade-offs. It surfaces that customers may be overspending on models when cheaper alternatives deliver the same results.
SEO meta description
David Totten, VP of Field Engineering at Vercel, reveals how a $9.3B company runs AI-driven GTM. AI agents eliminate CRM work, surface churn early, and automate customer expansion—without scaling headcount.
Target keywords
AI agents GTM operations
Vercel field engineering
David Totten
sentiment analysis CRM automation
consumption-based SaaS expansion
churn prevention with telemetry
Eve agent framework
AI gateway foundation models
customer-driven product development
sales enablement automation
post-sale expansion strategy
revenue operations AI
Read the complete conversation
Auto-generated from the episode audio and lightly cleaned. Expand to read or search inside.
Read the full transcript · 76 KB · David Totten
EDDIE0:04Welcome to Go to Market. In this podcast, we share tangible, actionable playbooks from the trenches working as go to market strategy and rev ops consultants for our clients here at Union Square. Consulting and candid conversations with revenue leaders in the market that have been there. Now let's get into it.
RACHAEL0:22Today I'm talking with David Totten, VP of Field Engineering at Vercel. Vercel is the platform. Much of the modern web is built and deployed on, such as companies like OpenAI and Adobe. David owns every technical touchpoint. A vessel customer has start to finish, and before this he spent almost 20 years at Microsoft doing a version of it at an enormous scale.
RACHAEL0:45Today we're getting into how Versailles technical work is built, the AI workflows they run from pre-sale to post-sale, and the results of his team's work. Thank you so much for joining me today, David.
DAVID0:56Fantastic. You can give me an intro anytime you want. I'll take. Yeah.
RACHAEL1:00Thank you, thank you. Awesome. So
RACHAEL1:03you're a VP of Field Theory. This is the first role of that type that we've ever actually had on the show. Can you give us a rundown of what that role does exactly?
DAVID1:12Yeah. Our job is to sort of display the art of possible with customers on what can be done with for sale, and give an opportunity to take advantage of the latest tech trends and tech innovation that exists. And so it's our job to meet with customers, big and small, help them figure out how to build, deploy, manage, govern, secure, integrate applications at any size or scale much easier, faster, and become more productive in their build and deploy lifecycle.
DAVID1:40So yeah, if customers on the pre-sale side obviously, and then obviously when customer shoes or so, which we have millions of deployments happening across the world, you know, how do we make sure that we optimize it and that they're getting the most out of the technology that that they have their fingertips? And then I also obviously manage all of our technical support engagement.
DAVID1:58If a customer gets stuck or gets hung up. We have a lot of self-service products within the stack itself, which I'm obviously going to talk about in this particular, but we also have humans in the loop to make sure that we can troubleshoot and go deep on any sort of configuration or optimization questions they might have.
RACHAEL2:15Yeah, and it seems like such an important, awesome rule of surprise that I haven't heard about it before. Maybe everyone listening to this is like, oh, you dummy. You know, it's such a common role. I've, I've been working at USC for, I think, three years now, and it's the first time I've heard of it, so that sounds awesome.
RACHAEL2:31And so where does field engineering sit relative to revenue generation and retention?
DAVID2:37Yeah. You bet. So if you think about go to market as an entity on its own, it's sales and marketing and technical implementation workloads. I own all the technical touches like you mentioned, with our customers. And so oftentimes we'll have accounting who really go deep with relationships, understand the business problems and outcomes that customers are trying to solve.
DAVID2:56And then it's my team's job to identify the right part of the solution or the stack that can offer to address those solutions. Right. And so there's field engineering orgs all over the world. I was blessed to to work for a founder at Databricks on our own field engineering solution over there, and chose to come over to Brazil because of the deep.
DAVID3:17I think tech innovation that happens here, as well as the fact that everybody's creating more applications and they need a quick way to do that that's super efficient and prioritizes the developer experience, but then also can kind of give you a harness and a control plane on all the agents and apps that are now being used in the development and production lifecycle.
RACHAEL3:36So walk me through a normal week. What are the what do you actually spending your time on day to day?
DAVID3:42It's a great question. I have a great team, I have a great staff. Were broken up into several divisions. Right. We've got a pre-sales, I've got a post sales deployment. I run a services business with our forward deployed engineers, where we go deep hands on with customers and help them build, deploy whatever they need. I own our technical support, so I spend a lot of time with my directs and just trying to understand what are the trends, what's working, what's not, what sort of cool use cases are we working with customers or partners on in design partners?
DAVID4:08We do a really good job of organizing feedback or customer obsessed, and so getting customer feedback and making sure we bring them into the development cycle of what do we want to produce next as a company, and we'll sort of technology or what do they leverage today that's a bottleneck that we can help make it more efficient gathering that feedback as well.
DAVID4:25So my day is almost crazy. It's, you know, hopping of customer conversations. It's dealing with technical escalations. It's being a beta private user on new things or product team launches, or it's just sitting down with our product officer and in our VP's of engineering and really working through real life scenarios that customers see value in, and where we can go and drive even deeper into the development lifecycle to produce products in the future that will help them.
RACHAEL4:55And when you first stepped into For sale. What did the technical side of the go to market or look like, and how has it changed since you've been there?
DAVID5:04Yeah. Great question. You know, fairly similar. We always want to talk and show customers demos and and have proof of concepts that we can launch and and have workshops. So we sit down and build with them and ideate on what's possible before I got here. Russell's known for just having super bright, intelligent, you know, people that will dive in anywhere.
DAVID5:24What we wanted to do was create a little bit more of automated engagement with customers so that they can find, try, deploy the tools on their own and serve as more of a true consulting firm to our customers that help them understand what they are taking advantage of, what they should be taking advantage of, or other bottlenecks or processes they have that we know we can help with.
DAVID5:45And so I've changed a few things, right. We created the forward deployed model as an example. We've now put a support engine agent at the front of our support infrastructure to make sure that customers can kind of solve their own problems, troubleshoot and solve directly within the agent of vercel itself, leveraging the tool to solve those problems. So they don't need to go back and forth over the phone or over email or slack or text message with a support agent.
DAVID6:12We've we've consolidated kind of the pre and the post sales. I really believe in helping customers see the vision of reversal can help and then stay with them through deployment optimization and the outcomes that they can drive. And so we're trying to shrink the swim lane a little bit on roles within field engineering. Just to simplify the fact that, hey, we have a relationship and a customer knows that they need help building or deploying applications.
DAVID6:35There's a technical person on the other end of the line that will help them, whether that's displaying and building a business case with them, whether that's building and launching a proof of concept, whether that's helping them build the demo they can leverage internally to their internal champions all the way through to, hey, I've got it deployed. Should I be taking advantage of other things?
DAVID6:52What other things come out as and orchestrate around? So I've really tried to streamline the process to make it simpler, more productive internally, but also for our customers just to know that whatever the case or the use case or the scenario they may have, they know they've got a technical person ever sell. That takes great pride, and frankly, it's compensated on their success with our platform, and we're going to help them.
DAVID7:15Whether the task is big or small, they'll have a solution that can address it.
RACHAEL7:18Awesome. And do you ever reach out yourself to customers if you're seeing, like, a signal that they're not doing well?
DAVID7:24Yeah, that's a that's a personal character flaw I have I just energy it's amazing. You know, we have an incredible product stack. But the stuff that customers are trying to do with or maybe want to do, but don't really know how to do it or have an existing process or workflow or tool that they're using that might be suboptimal.
DAVID7:43I'm just a glutton for those punishments. Like, I just am so inspired by the millions of exponential scenarios that customers have. And so I always believe one of my three leadership principles is proximity. I want to go deep. I want to go deep on tech. I'm going to go deep on existing tech that exists, that's out there that we might might not be powered by or sell.
DAVID8:02And I want to really understand where we can help and drive value with customers. So I like to go deep on our tech. I like to go deep with customer scenarios and use cases and really understand the outcomes that customers are trying to drive. So I meet with a ton of customers, and I find great energy and great aptitude for from our customers and our champions in these accounts that sometimes don't even know the full breadth of capacity or capability of that vessel can offer.
DAVID8:26But otherwise they're just really stuck, right? They've been trying to transform. They've been trying to get more optimized. They've been trying to balance quality and cost. They've been trying to drive a more productive developer experience so that they can, you know, retain their key talent, but also acquire new engineers and new product development teams. And, and I just find that these customer scenarios and use cases, they're all fairly unique.
DAVID8:47But at the end of the day, they're trying to make people faster, more efficient, like their jobs more, and really produce changes on the latest and greatest version of technology they can get their hands on as fast as possible. So, I mean, with a ton of customers, get a lot of feedback. And Vercel is very unique. I found in this space where we build, you know, 98% of the products off of a customer's use case.
DAVID9:11So we are very customer driven, use case oriented, where a customer has a really, really cool idea and we will go iterate and build on our product as a result of that conversation. So to a tee, we just came out of a really big marketing ship event where we launched over 50 new products and features and functions that will make our stack even stronger and really help our customers out.
DAVID9:33All of them start with a customer and a design partner on what are you trying to do and what do you do today, and how does it work, and how many steps and how long does it take? And what kind of internal taxes do you have as a customer on that workflow? We think we can help, and that's just a really cool perspective.
DAVID9:50We don't invest anything from For Sale because it's like oversell thinks this is a good idea. We kind of start with the customer, the workflow, the outcome first, and then we determine is that something that for sale can make more efficient for that customer. Right. And you'd be surprised. You make one thing work for one customer. That use case is generally repeatable in all other accounts, whether it's at industry, that segment, that technical challenge you're trying to solve.
DAVID10:14And so worry about these repeatable, you know, best practices in the product. And then it's my job to make sure that we we circle back with those customers or other customers who may take advantage of it and make sure that they're adopting it. And then we adjust, right. One of our core leadership principles as a company is iterate to greatness.
DAVID10:32So we'll build something with customer, and then we'll continue to fine tune it and train it and, and working into other products and features that we have to make sure it's applicable to every customer who probably faces a similar, if not identical solution.
RACHAEL10:46That's awesome. That's basically free R&D.
DAVID10:49I mean, we.
RACHAEL10:50We like.
DAVID10:52But it's like, yeah, if you've got users that report about the most.
DAVID10:56That are really giving you direct, honest feedback of why they've chosen something else or where they're really struggling in their workflows. Man, we just we just eat that up, right? We wrap something that we know will be adopted and have value and market and in our in our millions upon millions of community users, then just build something in a vacuum and hope people use it.
DAVID11:15And so I've worked at companies in the past that have built sort of solutions the other way, which is, hey, what do we need to do to defend our margin or build our moat? And coming to Vercel nine months ago. What a breath of fresh air where we do everything customer first. And what's the outcome that drive? And is that a space where we think we can add value?
DAVID11:36Then let's go build it and let's go deploy it and let's go build it with them to make it great. And then let's go rinse and repeat with other customers.
RACHAEL11:43Absolutely. What's been your favorite example of something that you've launched from? From that from a customer coming to you and being like, hey, I wish you guys had this.
DAVID11:55Yeah. We just launched something called Eve, which is our agent framework. Agents are all okay, you know, you can. Agents are everywhere. Now, why are we using AI? What are we doing? How do we do it? Like customers know, they need to be deep and leverage AI. They know they're missing out on some efficiency and productivity gains. Or how do I get to market faster with my own product, with my own supply chain, with my own financial records, whatever it might be?
DAVID12:16And we have a ton of really brilliant people who have built a ton of agents that we leverage internally. I'm sure we'll talk about some of them here on this, on this interview, but the fact that we have this honed in skill at building agents that make us internally better, make us our engagement with customers, stronger and faster.
DAVID12:34Customers want that, but they don't know where to start. And so when we just released Eve, it was fully focused on we have a lot of customers that know there's a way that could happen. I could build something to make this more efficient. I could be better. I could remove this text from my developers, or I could remove this text from product managers who don't have to spend as much time on this and can really focus on their core task.
DAVID12:55Where do I start? And so our framework is literally a push button framework to building your own agent, whatever you want on an internal app, an external app, internal workflow, external workflow. That framework provides the ability to build, deploy, manage, governance, integrate, secure any agent that you want to build within your company. And we basically have built that framework that harness around that process.
DAVID13:21So it's super easy to use. And now anybody can take a problem small or a big problem and put an agent behind it, quickly, put it into production, measure results and then iterate on their own within the framework. So Eve is, you know, top of mind for us because we just launched it and it's freaking awesome. And the number.
RACHAEL13:37Of that sounds really cool.
DAVID13:38Yeah. Or is that, like, our super smart people are probably building little agents here or don't really know the power of an agent over here. We want to help people build a framework and a structure and an organizing principle around that. And so that's been super powerful for us over the past couple of months.
RACHAEL13:56So Eve is an agent that builds agents.
DAVID13:59You think of it that way. Yeah. We can it's like this envelope of services that we packaged together that is super easy to engage with, prompt with plain text. Or if you're a developer via command line interface, and it can integrate automatically with services and your access control security, it gives you a governance layer on top of it.
DAVID14:20It gives you observability. So you can see what's happening on that workflow with that agent or with the applications that agent is interacting with. So yeah, it's kind of an on the agent on top of all the other agents that can possibly let you can leverage to build your own agent that's internal to you or externally facing food, interact with customers on like I gave you the example we're about, you know, we have a customer support agent.
DAVID14:44Anytime I go and I show people how the customer support tickets are managed, I mean, 90 plus percent of tickets are directly personalized and custom to a specific customer and their technical deployment of our product. Well, we have an agent in there that has solved a 90% of those inbound inquiries, right? Think about the economies of scale I can get from that.
DAVID15:03And we can help coach and optimize directly in the product, using this agent for customers to leverage and optimize their cell deployments. When you go to any customer who's like, gosh, I wish my support engine was smarter and faster, but I don't know where to start. Eve gives you framework that you connect to your own internal processes, your own documents, your own context, your own internal knowledge, your own security profile, your own identity integration.
DAVID15:27It can integrate with all of your databases, line of business, transactional, analytical. It'll do it all for you without having to recreate everything from scratch. With all those all those hooks and bells and whistles and provisioning services that you'd have to manually build before.
RACHAEL15:42That is so cool. I'm kind of jealous right now. I'm like, I wish you could have this for for our company. That'd be so cool. We use like, Claude quite a bit and we have, you know, kind of our own instances in Claude where we've had it learn a lot about our business to help us with certain things.
RACHAEL15:57But that sounds like that sounds like just the next level. Claude forgets all the time, and it's really annoying.
DAVID16:04Lacks context. And depending on the version of Claude that you're using and who's using that instance of Claude, you can have two different scenarios for the same workflow based on that. We use Claude a lot as well. Gigantic partners with anthropic love, Claude love Codex. We like cursor. There were partners with all of them. One of sale propositions is we are sort of tool LM foundation, model agnostic.
DAVID16:26So if you have a preference, great use it. But I'll just give you this simple data point that our CEO just continually brings up. And it just fascinates me. We are a deployment. We are a build developer experience deployment management, you know, platform and about 2% of our deployments. And we have millions of deployments on vessel, millions of projects, websites, applications get billed monthly with Vercel.
DAVID16:50Six months ago, 2% of the deployments were actually deployments built by agents were accessed by agents. Now it's over 50% of the deployments are actually agent built. And so that means somebody is going into Claude and they're trying to create an agent and they're trying to deploy it. And if you go to cloud and you say, hey, where should I deploy this?
DAVID17:09It's deployed over cell, and we provide the wrapper of services that give you all that infrastructure, you know, self-driving infrastructure, as we call it, that helps make sure that you've got a governance layer, a security layer, a management view, observability on top of it. But agents are everywhere right now. How do you build that cure integrated way? How do you make sure that you control access to the agents that get built?
DAVID17:31This is where customers need a lot of help. And so using the coding agent, we have our own prompt to design tool called V0 Claude cursor. Use any of these to deploy to Vercel and access that E framework helps make sure that you've got all those infrastructure pieces that are tightly integrated and secure for any, any sort of coding technology or tool that you want to leverage.
RACHAEL17:54Awesome. And so does for sale. Do you guys build all of your own agents, or do you also use other off the shelf tools for different parts of your go to market operations?
DAVID18:05Yeah, we've used a lot of off the shelf tools before for systems of record or for processing transactions. You know, we have our standard line of business apps, but more and more, we're leveraging our own in-house agents and applications to build. One of our principles, again, is you can just ship stuff. And that's an awesome thing for a startup and a tech based company to be able to go and just build whatever you want and deploy whatever you want, but you do need to make sure it's secure.
DAVID18:34You do need to make sure that me in field engineering, I'm not mucking up production level code on our core kernel. Right. So there are role based access controls within how you want to build and access these agents. We don't use any external agent builders beyond Vercel. We think it's the best, most seamless, most efficient, cost efficient, productive, quality agent building framework you can have.
DAVID18:54So we don't use any other agent builders per se, but we use a lot of coding tools, and we use a lot of line of business apps that we can directly integrate on top of, you know, for example, we just out in my customer support organization plane as our ticketing software. It's an early stage startup that's just a phenomenally fast and efficient ticket management system.
DAVID19:13In order to integrate that ticket management system, we have automation built by internal agents that prioritize the tickets and organize contexts with the tickets. And so we're building that with with plane to make sure that not only are we just using a ticketing system to sort of like organize, manage, you know, store the tickets over time, but we're building context into that UI.
DAVID19:36We're building context of customer sentiment into that product. We're proactively building product roadmaps or feature requests out of that ticketing system. And so it's how you interact with your own data, coupled with your and your own processes and your own design systems and your own, you know, interfaces that you have with all those applications and all those services, that's sort of the value of having a deployment framework like for sale, you know, integrate with those systems.
RACHAEL20:03And I want to I want to talk about your other, you know, tools and workflows and stuff in a second. But before we do that, I did want to I want to jump into what you just mentioned there, process and design stuff, because something we talk about all the time at you to see, like make sure your processes are, you know, consistent and predictable and reliable before you start adding on AI tools and automation or else it's it's garbage in, garbage out, and you're not going to get much help with your AI tools if you don't have the data to back it up.
RACHAEL20:30So what kind of or how did you figure out the right signals and data that you needed to input into, let's say, this customer support agent to make sure that it's surfacing the right kinds of tickets and triaging it properly.
DAVID20:46Yeah. You bet. I mean, again, we iterate to greatness. We think there's a better way that a customer, a technical support analyst, an engineer on my side can be more productive in a customer environment if they're served with the right context. Right. Like, you know, I'm a father of three boys. When one of them starts crying. The question is, what's wrong?
DAVID21:06Right? Are you hungry? Are you sad? Are you tired? Did you stub your toe? You have to get the context in order to address the problem. And so we start with feeding in as much context and real data about that customers environment as possible. And then from there you start picking up the patterns like we can sniff in the product itself.
DAVID21:24If somebody has a failure, for example, on their build, we can automatically fix that code within the vessel agent itself for that customer. Right. And then serve it to them to review it and approve it. Right. Because it's their code, it's their basis, it's their project, that type of power of serving more and more contextual information to the front end makes my people more productive by the time a human has to get in the loop.
DAVID21:48But more importantly, we can kind of serve. We know these problem areas. We know where the failures were. We're new where there are some questions or some potentially less elegant architectures that are being leveraged. We can provide those recommendations just based on the context of that deployment. And so if I know already that my kid is stubbed his toe and that's why he's crying, I'd go and get the ice pack before having to go and have a conversation with him.
DAVID22:12Right. The same way with trying to build an agent, putting as much context on our own personal experience, on the customer's environment and the tasks that they're trying to accomplish. All that context is super valuable, but it's heavy. There's a lot of different. There's, like I said, a million different exponential scenarios there. And so trying to serve up that right context and then build the signals off of that is what makes everything better.
DAVID22:35And by the way, we was something really, really important that I'm very proud about that Vercel does, I think, exceptionally well, which is we take all that data from customer support, we take all that data on where our failure is happening and where people have questions, and where are there issues with deployments or builds or access to a certain service.
DAVID22:53I can now organize that. So I actually have an agent that runs off of all of my support agent as well as in-person interactions, and now I can prioritize, hey, what are the top features and functions that we need to address, stabilize, put into market based on customer sentiment? Back to what I was talking about earlier, we're customer first, so if I can build these patterns now, I can build a product roadmap off of those patterns that will solve the most problems in the shortest amount of time.
DAVID23:20So building a park roadmap off of customer sentiment that I can run an agent and sentiment analysis on top of just makes our overall R&D process a lot more efficient as well. You know, we've experimented with a lot of agents just trying to make sure that people and the human element are as productive as humanly possible. Okay. And those agents, you get learnings and you apply agents into other spaces to then automate the next step or engage with the next set or synthesis of information or context.
DAVID23:50And so we just continue to iterate and build on top of each other the right workflows that we're trying to provide as much productivity improvement to the team as possible.
RACHAEL23:59And so what does that look like, that feedback loop that you've created, let's say like just to you looking at it on a dashboard, you have all these different AI agents running in the background. They're monitoring usage and they're monitoring issues and feedback and whatnot. Does that all get funneled to like like a dashboard that your salespeople and marketing people, whoever can just see.
DAVID24:21All the above. Right. So the marketing department gets what signals are working and what markets the EPD are engineering and product division. They get a stack ranked list of features or bugs and feature requests on a regular basis. It's ad hoc, but we also have a weekly dashboard, like you say, and a slack bot. That's all the top issues and features that we're running against.
DAVID24:42EPD actually runs an agent on their deployment framework on where they're pushing PRS and new things into product. That then goes back to my side, and we proactively communicate that out to customers. Hey, by the way, you brought this up six months ago. We've put it into market. I mean, I wish we had a six month dev cycle.
DAVID24:56Was like, you brought this up two weeks. We've already produced Marquardt. Right? Like we want to close the loop and make sure customers are aware that now we have this feature or function that they flag to us. So it's it's always this inverse. How do we actually service that information up with precision. It's one thing to just have a dashboard.
DAVID25:14It says here all the stuff that's happening. We want to say, here's all the stuff that's happening, here's how it's prioritized, here's the recommended set of actions, and here's an easy framework for you to click a button and send a message out to that customer, or turn this feature file on in that customer environment and tell them to go use it.
DAVID25:30Right. Just made that interchange of information a lot more accountable, frankly, between the two teams to take action on it. But also it's just made it a lot more relevant and precise on where we should we build, where should we market, which sales signals do we hear that we need to follow up with the human touch on? It's just helped prioritize the times, the team's time, a lot more efficiently.
RACHAEL25:53How hard is it to stand something like that up for maybe a revenue leader who's in a smaller company like 50 million to 200 million, something like that.
RACHAEL26:04You know, with the tools that exist out there today and like the agent building things that exist out there today, how does how does somebody else like, put that together when they don't have the amount of resources that for sale might have?
DAVID26:16Oh, man. Why? It used to be hard, you know, like I'm an old man now. So I've been in, you know, in applications and, and CI, CD and infrastructure provisioning and image optimization. I've been in this game for 25 years. It used to be really hard. And there's a reason why we have large DevOps teams at a bunch of these customers.
DAVID26:36A lot of the customers working with now, they don't have gigantic infrastructure teams or DevOps teams. They just can't manage it. So for sale as a core value prop is, hey, like self self-driving infrastructure, you just create a task or an application and we'll create all the plumbing and the infrastructure behind it to make it work. You want your developer capacity, which is always going to be strained no matter what.
DAVID26:57No matter where you are, you want your developer capacity focused on building a really cool product or a really cool experience for a customer, not dealing with infrastructure and connecting the dots. So it is very hard. That's the versatile value problem. We make it easy. And anybody up vercel deployment or an agent built and deployed through vessel. And it can do all these tasks that they don't even know as possible to be automated or gentrified, if you will.
DAVID27:21Okay. There's a reason the cost percent of our deployments are all on agents because people are trying things and they're testing new scenarios, right? We have a lot of production deployments, obviously, but one of the features that we really get a lot of traction on is a sandbox environment that you can take a live application and you can try anything you want with it.
DAVID27:39It's production code. I don't have to take a branch. I copy it over, I can try it and then I can see screen by screen. What does this element do? What does this agent do? What is this LLM or foundation model expose. What does it do if I run these other scenarios and other tools over here. And you can actually see performance and productivity wise in the application in the window without actually having to fork a branch, create a new PR, build a brand new application from scratch.
DAVID28:06And you can do this in a sandbox environment super, super efficiently. So we think that we think our bet and why I came to Brazil in the first place is, I think the people that create applications moving forward aren't going to be only developers, right? We want the developer experience to be super clean and fast, and we've done that.
DAVID28:23And we're going to keep building developer first, developer friendly infrastructure. But information workers, right, are creating apps. Teachers are creating. Advisors are creating simple dashboards for customers. Right. You know, front end and front line workers are going to be developing applications to do everyday tasks much more efficiently and have a track record of what's happening. And so the ability to create apps has never been a lower barrier.
DAVID28:49And you just have to make sure that you're doing it in a way that's controlled and secure, and then that somebody has a top down view on.
DAVID28:57What are all the apps and agents that are being built and what's happening with.
DAVID29:00Them. That observability layer that we provide across all your applications, I think is a really big benefit for any customer, however big or small you are.
RACHAEL29:09And what does the show look like on on that having all these different like agents, like fingers and every different department and feeding information back to one dashboard because I know AI is not perfect, it can hallucinate. And of course, if your data is not perfect, there's going to be issues down the line with AI as well. So how do you keep on top of that?
DAVID29:30Yeah, I mean internally and.
DAVID29:32Externally it's the same scenario. I just want to prompt I want to talk to something and say, I want an agent that does this right, our slack bot integration and work with Salesforce or even something like our integration with Shopify. Hey, I say, I'm a small business owner and I want to create a storefront. I should be able just to prompt and build a really slick UI that integrates with all of my systems and my commerce engine and everything else.
DAVID29:53But obviously I don't want to necessarily take that to production unless I test it. Okay. And so that's why for preview deployments, you don't have to push it to deploy to production. You get a secure URL that only you can access, and you can actually see what it looks like in an environment that would be pushed to production.
DAVID30:09But you don't have to push anything production until the human says, I like it, go right. There's a there's a big part of human in the loop for QA accuracy, privacy concerns. And so we want to be able to have the human at the front. What's the idea? What do you want to build? And then a human at the end, which is I love it.
DAVID30:27Let's go, let's go push into production and get it out to the masses. In between that, you want to automate and drive as much context, you know, through an agent as humanly possible to take out the work spent, you know, going back and forth on that. And so that QA process, where for sale is via a deployment, you push to production, and then once you're in production, you can immediately make updates and changes again, look at the show, value it, approve it with a human in the loop in the preview deployment before you push it to production.
RACHAEL30:55Quick pause. Everything we talk about on this show. Diagnosing go to market ops. Prioritizing projects for revenue. Impact processes. Metrics. Insights. Building a predictable go to market engine. We've built frameworks for all of it. They're free and undated on our website. Union square consulting frameworks. The link will also be in the show notes, so make sure you check that out.
RACHAEL31:18All right. Back to the episode.
RACHAEL31:20And so another part of your AI workflows that you mentioned in our pre call, you said your team doesn't do CRM data entry anymore. Yeah I thought that was really interesting. What what replaced it.
DAVID31:34Well I'm I got a bunch of engineers. Right. So if I go to my engineers I'm like hey I want you to open up your CRM, name your CRM, Salesforce Dynamics CRM, say whatever you want to do. There's a lot of good CRM systems out there. If I go to them and I say, hey, I need you to make sure you're spending an hour every day updating your opportunities or providing product feedback against each of these accounts and updating contact information.
DAVID31:56It's like, that's a tax, you know what I mean? Like if you're an engineer and a builder, you just want to go and work in solve problems with customers. Like that's a pretty heavy tax. So we have a process that if I have an engagement with a customer or I'm working on a project with a customer, I have an agent that runs sentiment analysis and details across that, and then I automatically push those updates.
DAVID32:15Super simple API that I've got the right permissions to go update that record in the CRM, and it updates it automatically. So I don't have to actually have that tax. The engineer gets to work with the customer or gets to work with whatever they're building and push it to production. And the agent picks up on those updates, picks up on dates, picks up on key contact information, picks up on product feedback.
DAVID32:36They may have been mentioned in the call, but never written down anywhere, and it automatically updates and organizes that in our system of record. The CRM. You know, we believe in CRM and think they're awesome. You have to have them. You have to have a system record for contacts and opportunities and accounts and and services and tickets that you have.
DAVID32:53I mean, every CRM is super valuable for any company, but the the manual task of filling out, you know, boxes on a UI like that, that shouldn't be a thing, right? We should automate as much of that as possible and then have the human approve it, like look at it, review it, and be like, oh, that's right. Oh yeah, I forgot they mentioned that product in that piece of feedback that down started written down for you.
DAVID33:16Push submit and it's all updated.
RACHAEL33:19That's awesome. I think that's one of the the trickiest things of getting, you know, adoption across your sales org for making changes to the process of how you are supposed to enter your fields and everything in CRM so that you have consistent, reliable, accurate data across everybody. It's like, oh, they don't want to spend, you know, half an hour after every call or I don't know, however long it takes to like think back through the call that they just had and check all those boxes and write in all those notes.
RACHAEL33:45Totally. Yeah.
DAVID33:46I think the chasm here is really interesting, right? If you're an individual contributor, you just want to solve problems. You want to work on cool stuff. You want to work with customer, you want to build cool things, cool products, cool engines. If you're a leader, you want to know that your people are working on cool stuff and they're solving problems and they're helping customers.
DAVID34:03But you also have accountabilities. What customers have you met with? What's our pipeline look like? What's our revenue forecast? How is consumption spiking on? What products have the biggest challenges? So like in order to get this Intel at a leadership level, to know how to prioritize resources differently in my space as an example, and where I need to focus more time in product or in actually like sales enablement or technical enablement or building new demos.
DAVID34:26Right? Like there's a chasm there on here's a person who does the work. How do I get the top line summary of insights to where I should prioritize insights differently? And there's a given and get there. And before it's always like a manager harping on his team, hey, fill out your data, take your notes, update CRM, you know, make sure you fill out the product or, you know, product change log.
DAVID34:45You know what I mean? Like there's just all this stuff. That stuff is a tax. If you're an icy who just wants to go and work. And so minimizing the tax on these workflows, I think gets me as a leader, the right data and insights I need to really judge. Are we progressing the right way? We need to with the right activities, with the right accounts?
DAVID35:04Are my people able to build and innovate with customers on the fly and really drive acquisition and consumption of our services? And it allows me to do that without having to ping somebody online and be like, hey, what happened in that? What happened in that sales call or what happened in this product evaluation? And what happened on that proof of concept.
DAVID35:21If the does that automatically, that chasm shortens right greatly. I don't feel the tax of having to chase an icy or a first line manager. The IC and the first line manager doesn't feel the tax of having to enter and update me for the sake of visibility, so I don't get fired, you know what I mean? Yeah, limiting that that that knowledge gap in an automated, prioritized way, that's just such a game changer for morale, for productivity.
DAVID35:48And also we can really hone in with real data and real insights where we should be focusing more time with our teams.
RACHAEL35:55Absolutely. I mean, it's the crux of what AI is always supposed to be, just freeing up all the time of like, all the things that a human doesn't have to do. Yeah. And making us more efficient.
DAVID36:06You raise a really good point on this, Rachel. I think we talked about it a little bit before, but, you know, AI can sometimes have like a bad connotation in terms of to take all our job, it's going to do all this other stuff. There are certain tasks that people actually don't want to do. You know what I mean?
DAVID36:20Like why automate that and let one focus on where their passion is, where their energy is, and the skill set they really they really want to drive. You know, it's like absolutely painful for an engineer who's working on building this great interface and integration with anthropic as an example. It's a really strong pain point to be like, okay, now you have to document everything that you did and how the service works.
DAVID36:43Like I guess as part of the job, but I just want to build it. I want to, I want to, I want to test and I want to try new things, and I want to toggle, and I want to come up with a cool way to think about this problem differently. Do I have to write that down? Do I have to put a doc around it?
DAVID36:55So if you can generate those docs and those FAQs and those how to guides and those integration script instruction sets, if you can document all of that automatically, where a person never has to actually sit down and start typing something, they can just go build, gosh, it eliminates so much of that noise, you know? And so we focus here that we think AI and agents obviously are going to be a big trend like everybody else.
DAVID37:18But it's less about job and individual displacement and more about individual productivity and making sure that they can shine in the space at a higher order of value than they were before. Taking a black belt ninja engineer and having them right how to document is not a valuable use of that resource. So how do I make sure that when that engineers plugged in and connected to this terminal and, and she's cranking out the next great innovation that a customer that thousands of customers are going to take advantage of, how do I make sure that that value added work is the main thing?
DAVID37:52Right. And we should figure out a way to automate so that you don't feel that tax, but it's still there because it's required so that other customers can find the solution and other people know how to talk about it. People can integrate with their projects. So again, it's like not about job displacement. It's really about value offering and value prioritization of what are people really good at and how do.
RACHAEL38:16We make it up. Yeah, especially when you have like these really high performing IQs that like they want to and they need to get into the flow of whatever they're doing and then having some kind of redundant, mundane tasks like logging something breaks that flow. And then, you know, it takes time. One of those people who needs to get into a flow.
RACHAEL38:35And if something comes up that takes me out of it, it takes time to get back into it, and then you're losing that time that adds up and yeah.
DAVID38:42Yeah, 100%. And just again, a simple agent on our sales calls with customers, we record that. We put it into a sentiment analysis agent. It tells us what was asked how did the person handle it? Was it good, bad, ugly? What are the next steps? By capturing all that, providing sentiment analysis and then feeding that into various engines.
DAVID39:04An enablement calendar, for example. Hey, this person to this proof of concept template that we leverage or push them to our central demo zone for the next time. Anything that we can do to automate that process of writing it down, or having a manager sit on the call for an hour to really get a gauge on, do they know what they're doing or not?
DAVID39:22We don't need to do that anymore, right? We can just use this aging to give us the feedback and the sentiment analysis. And then where do we need to prioritize resources, whether it's coaching, training, better product, better pricing consideration, better contract language, we can automate all sites without a major to sit on the call and release and be like, are you talking to this customer?
DAVID39:41And what were the next steps?
RACHAEL39:42Like, yeah.
DAVID39:43Yeah, let's talk to the customer.
RACHAEL39:44As unnecessary.
DAVID39:45And then after the fact. We'll circle back with all that intent and insight and really have a firm and focus set of go do's that the team can then go execute against. Yeah.
RACHAEL39:56I wanted to switch gears a little bit here and talk about post sales and expansion. So the post sale, you know, it gets talked about far less than pipeline and pipeline generation for most businesses. But for a business like For sale that runs on a consumption based business model, primarily it's it's most of the revenue expansion and retention.
RACHAEL40:18So how do you think about the post sale motion at Vercel, and what tools and workflows have you deployed to help with that?
DAVID40:28Yeah, so a couple of things. This could be just like a philosophical shift, but I just really don't believe in like pre-sales versus post sales to a certain extent. Let's say you get a contract for a specific use case with a specific set of workplace. Yeah. That's great. We show a value, we show a prefix the customer wants to buy from us.
DAVID40:49Once they start using vessel, there's probably 18 other workloads and probably 100 other use cases that they could use that product for that set of services for, or other services that they don't even know about. So I kind of believe that everybody is in post sales to a certain extent, because vercel developer friendly, most developers in the world know about it, use it all the coding agents deployed over sales.
DAVID41:11So it's like if you're building something you've probably heard of or so you might use it in in your personal life and your hobby situation. So it's really more about like, are you taking advantage of all the services that Purcell has to offer? And are we simplifying your development and management process across applications and agents? Okay. So I kind of think of everything is kind of like, oh, sales number one.
DAVID41:32And then number two, the expansion motion is really like telemetry on the customer. What they're doing, their feedback and how they're using the product in product. And then how do we give them tips and tricks to take advantage of those things. You know, the the classic example of if somebody is using a foundation model, which we love, people have gone out and spend a lot of money on OpenAI and Anthropic and Gemini, you know, name your foundation model and they've gone out and spent it.
DAVID41:57We love that. We want people to experiment with agents and tokens. But if you just ask the question, why did you choose that model? Why did you choose that lm y for that specific task? Typically people don't know. People are kind of like, well, the most popular one or my brother in law's sales rep. I just I'm going to buy this and I'm going to go do it right, like you were talking about quarterly.
DAVID42:15It's just what we use. And these tools are getting so much better. But they kind of leapfrog each other every once in a while. Like every two weeks. Every two months. They're sort of a different new flavor of the week that you can go and leverage. One of our biggest value propositions is something like AI gateway, which most customers aren't even using right now.
DAVID42:31And the question is why? What AI gateway does is it lets you service up whatever tokens you want to use into a through a gateway. You can leverage tokens you've already purchased, but also obviously you can buy new tokens and you can outline the use case or the application or the agent you want to build, and so will help you through that gateway.
DAVID42:48Tell you what's the right LLM Foundation model to use for your cost and your quality trade off. That's a brilliant value proposition, right? For anybody is like, why am I spending all this money on this model when I can do that same task at the same quality for a 10th of the cost? And that's a real that's a real use case that a bunch of Versailles customers aren't even using.
DAVID43:08They're using us to play and build applications and service it. But then they might have another group who owns the token infrastructure and how we're deploying and using LMS internally. So when I think about post sales, which is every customer, it's, hey, are you taking advantage of everything that verso has to offer? And then the second piece is even within Roselle, there's always new products, new features, new optimization scenarios that we think we can help customers with.
DAVID43:34So I spend a lot of time looking at random, like usage spikes up or usage spikes down, or how many customers are trying new workloads this morning. And I wonder what they're doing, which customers aren't using AI gateway? Can I call them and ask, why do you not know about it? Or are you hard keyed into whatever foundation model you might already have?
DAVID43:53That telemetry that we have after a customer starts using the services that the customer has direct access to, they have a health dashboard, they have their observability report, they can track their costs and their spend against each of these workloads. And then they can ask the agent, hey, I'm spending way too much over here. Am I doing something wrong?
DAVID44:11Could I am I my architecture be set up differently to optimize the spend here? And the agent will tell you itself, oh, you're not using this, or you should be using this, or you're using way too much of this on a feature function that Purcell offers that could be done differently, right? We're very price sensitive with our customers.
DAVID44:29Yeah. And we don't ever want to be in the position of, oh, we love Vercel, but it just costs too much. There are cool, awesome innovations that we have that are less expensive, and probably for the tasks you need, you don't need all of this heavy duty infrastructure to support it. You can use some layer weight services from Brazil and make sure you take advantage of it for the tasks that you need without spending all that money.
DAVID44:50And we flitch, we flip to something called fluid compute. We only charge when people are actually using compute scenarios. Oftentimes, when you're working directly on the hyperscale environment, you're paying for compute no matter what, doesn't matter if you're previewing, building, deploying in a production application, it doesn't matter for us. You only use a compute for when you're actually using and building and deploying the application.
DAVID45:10And so we have these little tips and tricks with observability tools that we can show the customer, like, hey, you're spending a lot on your infrastructure servicing. You know, Versailles has this thing over here that you could probably cut that in a 10th of the cost and get more functionality out of it. Just think about experimented with and here's how to go do that.
DAVID45:28We can serve all that up in product directly with customers, and they can choose their own, their own experiences to go do. I mean, the reality is so many new products and features like I don't I don't have like a standard play, I can go to my team and say, okay, everybody go talk to your customers about AI gateway.
DAVID45:46You know, they say, yeah, there's like every customer is unique. Everybody's using the technology and the Swedish service is differently or changeable. So it's kind of like, which customer am I really intrigued with? In what use case do I really want to get behind and see what they're doing? And then from there, we can sort of build out ideas and art of possible conversations with those customers.
RACHAEL46:07And so what's your churn prevention play then, when maybe you see signals that are warning signs that you've come to recognize?
DAVID46:16Yeah, I mean, it's a very fair question because like.
DAVID46:18Churn is a real thing and it's very concerning to us because we believe that churn happens not because people are like antiviral or like, oh, I just hate the triangle company. You know, I was like.
DAVID46:30Because.
DAVID46:31They don't have optimal architecture set up. They're not listening to the best practices or the agent provider recommendations that are there. They're not taking advantage of the services stitched together the way that makes the most sense as a full platform, right? Most of the comes because they're using one aspect of the product, and they're not realizing, oh, you're just too slow or too fast or too expensive because you're not using the full platform and you're using that elsewhere.
DAVID46:55And so a term prevention is a big thing. Again, we flag for people that are spending through their commits faster than they should be. We also obviously manage and look at observability on what's our steady run rate, what we think is an appropriate growth rate for these customers. And there's something changes. What's happening? What are you doing? Can we help you with something?
DAVID47:13It looks like you turn something off or you turn something on and your costs are gone up and your workloads have gone up, or your token usage has gone up. How can we help make sure you're being most efficient? Right. The churn here is fascinating. It's it's actually quite low in the pro and the enterprise space is that it's actually quite low.
DAVID47:34And that's because the product just works too. We drive a lot of the observable and direct recommendations in product to customers. And then three, it is like really easy to narrate trade offs within the product. And I think that that's fair. We don't want bad revenue. We don't want customers spending a bunch of money with us for bad architecture and bad practices, like some churn is by design, right?
DAVID47:56You just have a bad pattern and you're asking for sale to do something that's really expensive with us that you could do elsewhere for some other task. But our churn is actually fairly low, but we are sensitive to it, right? We want to make sure you optimize the vessel experience and use it for what's really, really good at and don't get caught trying to use it for something that they shouldn't be using it for in the first place, and they'd be super disappointed with the outcome.
RACHAEL48:21Well, a theme that I'm recognizing is absolute visibility, both on the customer use side so they can get whatever information that they need through chat agents stuff, but also on your side and your team side. You're talking about like you can just scroll through dashboards or look through, you know, this data that you're getting and you're able to see everything that's happening from top to bottom and be able to see, like what, what customers are doing in their platform, you know, to, to to notice signals.
RACHAEL48:53That's something needs to change or you need to step in or do something. And I think that's really awesome because visibility is is chronically the number one thing that crows and revenue leaders don't have enough of. And, you know, the exits don't have enough of like how do we get more visibility? So how did you guys build the foundation of this complete visibility, where the data is so readily, readily available for everyone and it's trustworthy and accurate?
DAVID49:21Yeah. So we know high level ops. Absolutely. Where people are using certain services and where there's spikes or where there's big drops or where they think something's bad. And we're like, oh wow, they're doing something. But because every project and customer is private and unique, I can't I don't know what function is. I don't know what code it is.
DAVID49:38I can't see their IP. So I so it's really like while we have deep levels of telemetry because of a web based service, and we can sort of measure the product performance, I don't know what their tenant is doing or what that project is. I don't know is that for an e-commerce website or is that for a marketing website?
DAVID49:54We don't know. You know, and we do that intentionally to make sure customer data is private. We do that intentionally. Yeah, like building new products that will compete with somewhere or their IP or whatever that is. Like we don't want any of that business. And so it's really my line where we will see the telemetry in big buckets, but then it's up to our relationship with the customer.
DAVID50:14You'd be like, hey, we noticed this on our end. We don't have any idea what it's from. We don't know, you know, which project or which kernel or which IP customer facing internal. We know we don't know that, but it's our job to make you aware. Are you seeing this too? And some best practices around this spike or how to optimize this differently would look like this.
DAVID50:32And then you go solve it yourself in the product with the agent and come back to us if you have questions. And again, trust the platform for a decade now. And so customers nine times out of ten open the Komodo with us. Right. Like, oh hey, this is it. This is the code block we have. We have more customers than I prefer trying to share their GitHub repos with us and trying to share their code.
DAVID50:52And I'm like, no, no, we can't take that. You know, that's I can't do that. You you keep your IP but but we'll help walk you through how to troubleshoot some of these calls or functions or workflows or, or queuing, you know, strategies to make sure you're optimized for performance against cost. Because customers are very trusting of us for a good reason.
DAVID51:10And we've earned it. And, and, and customers know we have no ill intent. We're not going to replace their product. We're going to steal their IP. We're not going to go and build our own thing that does what they do. Our job is to connect the world's best applications and integrations with each other and make that process simpler, faster, cost efficient for anybody who wants to do that.
DAVID51:30So yeah, I'm lying on telemetry. It's funny because, you know, when I first, first got here, there was a lot of data that we probably had access to, but nobody really looked at. And it was kind of like, well, even if we had that data, we looked at it. We really know what to do with it. You know, I don't I can go to a CIO and be like, hey, this function and this call for this piece of IP is growing, this error.
DAVID51:53I don't I don't know that I can't see that it's really up to the customer to be able to troubleshoot on their own. But I can see, oh, you may have spiked something in an aggregate level. Y'all got to look into this. And here's the spike data that we see. Go into your own dashboard. Look under admin. See which applications or functions are spiking.
DAVID52:11You can actually just ask the agent within Brazil to do that work for you. I can't do that for you, but you can let me know if you need help troubleshooting. And I actually think that keeps us from nerding out and going to deep on every single customer deployment or project or or application. It keeps us out of the weeds to a certain extent on just totally going deep and and line by line code.
DAVID52:33And it also puts onus on the customer to solve their own problems. Right. And, and really optimize their trade offs on their own as opposed to us doing it for them.
RACHAEL52:42Yeah. Do you ever have issues where a client is having difficulty seeing results or progress with Vercel, but it's because their underlying data is not great. They don't have the right processes in place. And you know, obviously it's not going to work out.
DAVID53:01Yeah, 100% of the time.
RACHAEL53:03100%.
DAVID53:04Well, we could.
DAVID53:05Buy me a clean data.
DAVID53:06Set that people can build applications off and trust and know it's always going to be correct and accurate and surface the wrong way. Contextualized. I'll buy you a boat. You know, I just I don't know that that exists. The funny thing about data is that it's very subjective quality. Data quality is very subjective. So what one person might think is the right data elements to gather and organize and provide context way, another person might see it exactly the opposite.
DAVID53:34Okay. It's the same thing as like sales cares about accounts and forecasts and quota attainment. Finance cares about how much money did we actually collect and is it good revenue or bad revenue. Right. Those are two different views on the same data. Yeah, I got a customer. I got a dollar amount. I'm consuming stuff. Finance. And our leadership team looks at it totally different, which is like are they using the services the right way?
DAVID53:57Are they optimized against our services? Are they getting the value and the outcomes that they want out of that revenue? It's the same data set, but we look at it totally different ways. And again, I think, you know, there was a trend five years ago that started about, gosh, probably ten years ago. Data just keeps coming in. It's more and more data.
DAVID54:12So how do I organize it? How would I grab it? How do I build the semantic layer on top of it to make it even more effective with agents? That's going to be a journey every customer is on, right? Every customer. I keep figuring this out, but yeah, to a tee, people build applications and agents that are awesome, but maybe the underlying data is not as reliable, or it doesn't get updated frequently enough, or it's organized in a way that kind of hallucinates that agent or AI on top.
DAVID54:37And so that's the value of being able to build quickly, deploy even faster, iterate real time based on what you believe or do not believe. The data is actually organizing or orchestrating you toward. Right? You know, if everything was perfect on the back end and you could just run an application inherently forever, and it always gives you greatest, best in time results.
DAVID54:58You wouldn't need any developers, ever. It's kind of like I could just like, push a button and I just run this dashboard forever, and I know it's always going to be right. Business is change. Lines of business change. Data elements you're capturing get organized differently. The key to staying on top, you know, what's that saying? You know the only constant is that there will always be change.
DAVID55:19And so yeah how do you iterate builds value. You know, QA these applications on top of an ever changing landscape of infrastructure. You want to do that fast and and securely, right? You want to have security. You want to have your environment be your environment. You want to share that with your competitors or a platform company or an icy or a hyperscale, or you want to be able to to control that and iterate quickly on it.
RACHAEL55:43That's awesome. It's like we probably even just team up a little bit because you'll see exactly what we do is helping companies get their data foundations and their data process foundations in a place where they can actually use AI and use it properly and effectively.
DAVID55:59Well.
DAVID55:59There's three layers, right?
DAVID56:01There's like.
RACHAEL56:01Yeah.
DAVID56:02The core system and data infrastructure. Like that's one layer. The middleware of today is business process workflow capture what's valuable, what's not, what's secure, what's private, what's not. And then the top the application layer. Right. Like how do I access it? How do I organize it. How do I update it. There's still that application layer. So those three layers have been the reference architecture of software development since the beginning of time.
DAVID56:26Those are still super relevant. And we're trying to make sure that the top layer of application and agent builds, as well as the middleware layer, we're trying to make sure that those, you know, access and connect with the data layer in a way that makes the most sense and actually gives you insights on whether or not it is a quality.
DAVID56:44Is there is it right? Yeah. Is it fit for purpose of what you're trying to do? Right. But every customer I tell with the day layer, that's just it's a it's a foregone conclusion how you make that data layer or that data lake or that database useful for somebody to take a specific task on top of. We want to accelerate that process, even if the data is in a hip hop place, depending on who you ask,
RACHAEL57:08And it's a good thing it's coming from because that's why we're in business.
DAVID57:11Well, I think so. I think that's 100% correct. Yeah. We have job security knowing. Right.
RACHAEL57:15So yeah. Yeah.
RACHAEL57:16what are you working on next at Roselle? What's what's the the version of your Aug 12 months from now.
DAVID57:23Yeah.
RACHAEL57:23That you can actually talk about.
DAVID57:25It's a really fair product. So my boss Jean grocer is amazing. She's a she's an absolute unit. And she has beat this into my head that we treat GTM like a product, just like you would an engineering team. So we want to do sentiment analysis. We want to analyze sales trends. We want to see who's good at landing and succeeding and proof of concepts and demos.
DAVID57:46We want to see people who are contributing to the core of or sell product itself. So how do you organize all those things, and then how do you make sure that you are building a proactive engine out of it? Okay, so I would say where I started for sale, the organization was probably like 80% reactive. Oh, this came out.
DAVID58:02Let me see what I can do with it. Or the customer has this question. Let me see if I can go solve it. I'm trying to build a proactive, like, I want you to wake up every day and be like, oh, cool, I'm inspired by something. Let me go reach out to these customers. Well, let me go work on this product.
DAVID58:14Let me go build this demo that I think will hunt. And so my organization, you know, 12 months from now is going to be all signal driven, all proactive engagement, all what if I told you I could do this, that and the other thing, right? Or what if I told you I had this idea and I could build it in less than two hours because I use for sale, right.
DAVID58:31Like, I want it to be super proactive and we are proactive. And what you get to do in anything in life, whether it's work, personal relationships, your parenting philosophy, it doesn't matter when you're proactive. You feel in control. You feel empowered. You feel really helpful. You don't feel stressed or underwater or consumed with all the inbox and slack messages you got to get to, you feel like you own what you're doing every day.
DAVID58:55And so, like my dream, I always say is I want a few of engineering to be this place where you've got empowered, you know, black belt ninjas that are just going out and just are so passionate about the problems they know they can solve and they can do it on their own timeline and they can do it on their own empathy.
DAVID59:11They can do it on their own priorities.
DAVID59:13We're going to give them a bunch of data. We're going to organize a forum. We're going to give them insights. We're gonna give them best practices. We're going to give.
DAVID59:18Them hunting grounds, but we want them to own where they spend their time. And I have found that if you have an individual or a leader who really feels empowered to own their own journey. They also then feel accountable to the results, so one enables the other. I have not found that accountability creates enablement and empowerment. I haven't found that okay.
DAVID59:40I have found that Power man lands and makes people part of being accountable to the results and what they're supposed to be doing every day to add value to customers.
RACHAEL59:48Yeah, it's really a leadership philosophy too.
DAVID59:51So good. It's worked for me.
RACHAEL59:52So yeah.
DAVID59:54Yeah.
RACHAEL59:55Awesome. Well, thank you so much, David. This is a really, really insightful conversation. Where can people go to follow you if they want to see what you've been doing or check out for sale?
DAVID60:06They find me on LinkedIn at David W Totten, David W Totten. All one word but also for sale is a phenomenal yeah, you know LinkedIn Twitter, everything else to go follow for sale. Com if you want to get started go today and then our CEO g r a UK is a phenomenal Twitter LinkedIn follower. This guy is on top of every tech trend, every possible scenario, and he's openly sharing best practices.
DAVID60:34And he's open sourcing a bunch of our projects and our tools and our skills that are out there. Mass op for the mass public to take advantage of. So I definitely recommend going and following G.
RACHAEL60:44Awesome. I'm going to do that too. So I love a good innovative tech. LinkedIn thought leader.
DAVID60:51Was the best. He's the best. He's incredible. So and.
DAVID60:53He's.
DAVID60:53He's he's kind of funny right? A tech startup who's got strong opinions. But he's right 98% of the time.
DAVID60:58He's so tied in with all these other tech founders and all these other solutions, like the threads that they can get on together is really inspiring. And shows like The future, we're in this amazing opportunity right now with technology, the future of what's going to be at our fingertips that we can use and deploy and access is just we're at the we're at the tip of the iceberg.
DAVID61:15And so to follow these tech founders and their true visionaries on their vision for what the art of possible is, it's my favorite thing. It's like I doom scroll on.
DAVID61:25On our CTO CEOs.
DAVID61:27You know, Twitter accounts.
DAVID61:28I'm like, well, what's he up to now? Was he talking to you? I don't even know what that company is. What do they do? It's pretty cool.
DAVID61:33It's pretty cool.
RACHAEL61:34That's awesome. Well, thank you so much, David, and I hope you have a great rest of your day.
DAVID61:39Rachel. You're fabulous. What a blast and I'll see you soon.
EDDIE61:43Thanks for listening to the episode. If this resonated. Please give us a five star rating and a follow. It helps us reach more people, and you get our latest and greatest content without having to search for it. And if you're looking for hands on help and go to market strategy and or rev ops, please reach out to us.
EDDIE61:57We help our clients with everything from annual planning to improving processes and go to market, implementing systems to support those processes and go to market AI. We're always happy to offer a free consultation to help you identify the best opportunities to improve your go to market engine, with or without our help. You can find us at Union Square consulting.com and the info will be in our show notes.
