The Future of Software Development with Ahmed Bashir (DevRev)

Mark: 00:45

Hello everyone and welcome back to our next episode of the CTO Compass podcast. Today we're joined by Ahmed Bashir, the head of engineering at DevRev and I'm really excited to have him here. DevRev is an amazing company. I've worked with them for about a year now and very happy to have Ahmed here. Ahmed, could you tell us a bit about yourself and just a quick intro?

Ahmed: 01:07

Yeah, first of all, Mark, it's a pleasure to be here. And thanks again for setting this up. My background, I've been at DevRev leading engineering since 2021. Prior to that, I spent almost 13 years at Apple, was part of the founding team that brought us iCloud and iMessage. In the same year, in fact, and subsequent to that, had a role to play sort of leading aspects of push notifications, FaceTime and other technologies as well.

hnologies at the time. And in: 2021

So that's my background.

Mark: 02:07

Cool. That's really exciting.

So I love that you worked for Apple, of course, and you're right, that was one of the forming periods of technology at Apple and the internet and cloud and everything. That was so cool.

So why did you join DevRev? I think you're based in California. You could have chosen probably a dozen or more companies. What was exciting to you so much about DevRev?

Ahmed: 02:30

Yeah, I think the space was really exciting at the time. The idea of being able to reimagine work, obviously, you know, we had seen a transformation over the early noughts and the 2010s where companies were starting to think about going from their own core technology, maybe a shack or a small data center that they were operating for their businesses under the CIO. The CFO is probably always talking about what it costs and why things are going down and, you know, being able to run core business practices with their own infrastructure and then being able to suddenly elevate that to something that scales to infinity with the cloud, with, you know, the highest quality technologies available to everybody. And it was a pay as you go model. It was very exciting. And I thought that at one point, We had sort of reached the potential of cloud and it just became a game of convergence, really a game of adoption beyond a certain point. And what excited me was just the space of being able to reimagine what the next level of innovation looks like in business. And a lot of times what happens in business software is that we start to see what the best of the best looks like in consumer software and start to try to envision what that would look like in business. And that's what happened when the iPhone came into business, right, where people had iPhones at home and they just had to buy a new one. I did, you know what, why am I using this pager at work? Why am I using this BlackBerry at work when, you know, I've got obviously better technology on my home front. And so the same thing would happen here where I felt like as. Technologies like large language models, and just being able to convert information and answer questions and build more and more degrees of automation became more apparent in our personal lives. It would definitely start to reimagine itself in the business world as well.

So it was really the space. And then of course, when you're going into a big space and you wanna do something where you're truly reimagining how work gets done. You want to partner with people who have done really big things and have an innovation sense and taste and leadership and everything else. And, you know, I think I had prior to my joining Apple, I had met with Dheeraj by happenstance.

nity that presented itself in: 2021

Mark: 05:16

It's interesting. So I've met Dheeraj and indeed, he's one of the most amazing founders that I've ever met.

uld follow him. Still, it was: 2021

Ahmed: 05:44

Yeah, Mark, you know, when you start a company, especially when you're starting a company, it's less about whether you think it's just going to happen. And it's more about, you know, just having that sense of confidence and your ability to actually make it happen. And so you're right, it was, you know, significant, I would say probably 15, 18 months before things really got cooking on the GPT and OpenAI front. We obviously had a lot of confidence that these elements were coming together. The idea of transformer models was out there. People are already starting to talk about it in close circles. And obviously, people understood the value of models, even if they weren't the models of today, there were models that were being distributed and there was federated learning and there was privacy around models and there was a lot of A/B testing. And yes, it was probably something that machine learning experts and data scientists were tinkering with, as opposed to sort of the general engineering audience. But these technologies were starting to make themselves known. And people like myself and others You know, we were paying attention to this.

So yes, it was early, but we felt like there was enough there. So that's kind of how it all came together.

Mark: 06:57

Okay, that makes sense. It's quite interesting.

So the question then is, I think not everybody that's listening knows what DevRev is. Can you tell me a bit about the actual platform, the application and what it does?

Ahmed: 07:11

Yeah, I mean, DevRev is, it's a way to reimagine how work gets done in a way that connects your team intelligence, the information that's driving the productivity within your enterprise, and it kind of brings it to the fore and allows everybody to sort of benefit from that information. So we talk about democratizing team intelligence and the way we try to manifest that in to create value for businesses is through search analytics and workflows.

So, first thing we do is we try to effectively bring all your critical data and integrations and make those available to you in all the core surfaces that you have in your business. We try to make sure that you can actually take action on the information that you have, not just read it, but take action on it. And by doing so, we basically change the way work gets done. We bring departments together. We get you to be closer to your product and closer to your customer.

Mark: 08:06

Okay, that's really cool. I've seen the platform and I know what it can do and it's incredible. Exactly.

So the next question then is, and I think I've heard you or I'm annoyed to hear D-Rod say this quite a bit, DevRev is a platform that didn't get AI bolted on later, which a lot of platforms do. It's actually a platform, and I think you said it before, that has AI built in from day one. Why is that different and what makes that different for DevRev?

Ahmed: 08:35

Yeah, it's true. Mark, when I talked about search analytics and workflows, the reason we talk about those in the same breath as AI is because we think that that's how most knowledge workers work. Will practically benefit from AI. It's not about doing things that you've never done before with AI, it's about doing things in a way that you haven't done them before. But you're doing similar things.

Like for example, if people are trying to discover knowledge or search for information, typically speaking, they would have maybe done a keyword search or perhaps something of that sort, or perhaps they would have tried to bookmark something or save something, right? And now what we're saying is that you can conversationally access information and then basically take an action on it. And that action can be described very conversational. And that's what makes it different. That's what makes it intelligent in this sort of new way. Similarly, with analytics, obviously, we've had just a whole slew of really interesting analytics tools that have come out over the last 10, 15 years, but now they become increasingly conversational, and they become available to you to take action on, which makes it really valuable.

And then, of course, with workflows, we've had workflows for decades. Decades, but now rather than being very deterministic and conditional, you have These systems that can sort of connect with sensibility nodes.

So sensibility comes in the form of the types of things that we used to do. Clustering, summarization, prioritization, classification. These are things that typically humans would do. And now with the ability that large language models have to understand language so well, that they can do all of these things relatively well, classification, prioritization, summarization. These are things that, and I'm sure most of the audience has tried it. They've given large amounts of information to these systems and said, "Hey, can you just give me the gist?" And it does an incredible job, and it does it incredibly quickly. And so you take advantage of that by saying, listen, my automations aren't going to be the way they were. They're now going to both kind of sort of connect and chain together deterministic behaviors like we had before. But now instead of escalating or having to have sort of these stoppages where a human decision maker has to sort of add a little bit more information so that you can carry on with your automation. You can actually connect and chain together a series of automations with these sensibility nodes. And that's what makes these new workflows and these new automations so valuable is the fact that they can actually reduce your time to resolution, they can actually reduce your cost of action, because all these things that used to happen in between automations, through humans can now continue through machines. And so now workflows and automations look very different than they used to. And now you're going to have these automations carry on for longer because they're doing more and more complex things. And they're starting to do things that typically humans would do, but at scale, they will start to become difficult to do consistently.

Mark: 11:23

Okay, so where you used to have humans in the middle of workflows, basically those steps are going to be taken out by AI. Like DevRev. It just becomes one long workflow instead.

Ahmed: 11:35

Yeah, I think longer workflows is what I would say. And sometimes that has to do with rationalizing information. And, you know, sometimes, you know, if you look at the comparison with a human doing those specific tasks, as these tasks become more and more procedural, onerous even, you know, even our ability to sort of stay focused on the task and to do it equally well every time gets compromised. And that's just the way it is. Usually when you're doing things and it's very, procedural, we tend to get distracted. We tend to do fewer actions per hour over time because we just get burdened by the monotony. And, you know, for us to be able to exchange that work for more deep work.

So, for example, for me to be able to maybe reduce the number of times I'm looking at a new enhancement that came into our system and to just respond to the person by saying, listen, the outcomes that you're trying to describe in this idea, in this enhancement, in this initiative, they're not very clear. And what I mean, they're not clear. What I mean to say is that the done condition is not clear. What you want to release, what the benefit to the user is, those things are not clear.

So in my line of work as sort of a head of engineering, I may actually have to review and triage information, ideas, opportunities that are coming into the system. And one of the first things that I would do is make sure that it's a complete thought. And oftentimes when I'm relaxed and I've got some free time, I'm going to be able to I will do a good job of sort of taking that information and providing some clear feedback to the person who created the idea or the initiative or the enhancement and say, look, this is what you need to do. You need to clarify the done condition. You need to clarify who the audience would be. You need to understand who the buyer would be, so on and so forth. Now what I can do is I can characterize that in an automation and just make sure that anytime a new idea comes in, this is sort of the feedback that goes back to the. The person who created the idea and they're able to revise it. And now by the time I look at it, I already know that it's gone through a certain degree of clarification. It just makes my job easier. Now, am I not doing that anymore?

Well, of course I still do it, but I do it at a, I meet the idea at a place where it's just way more refined and it just helps me be a little more clairvoyant. It helps me sort of. Maybe contribute at a higher level than perhaps those basic steps that I have to take repeatedly. And just imagine in a quarter for a company like ours, where we have hundreds of engineers, we've got 600 plus people, just imagine how many of these ideas are coming in, my ability to actually help at the earliest stages, the mid stages, the late stages, you have to believe that as these dozens and dozens of ideas come in every month or every order, my ability to be consistently good at rationalizing what's happening and being able to provide really good beneficial content back to the person so that we can uplift the idea. I still have to do everything else in the mid stage and the late stages of that idea if we continue to execute on it. And of course, being able to get that additional help in the early stages, it just helps me that much more to be sort of available to them in the mid and late stages, because there's only so much time and energy that I can provide. And what I do is I find myself more sort of, I would say, uniquely sort of. Committed and maybe specialized to help in some of those later stages versus just providing some of those remedial sort of tasks that are maybe very basic feedback back to the person who started. And this is just one example. I can give you dozens of examples where similar actions where I'm able to sort of take remedial tasks and provide feedback to people and decentralize that effort of refining and revising ideas so that they get better or maybe even providing just so Little checklist items, but not just in a way that's incredibly passive, but in a way that's very active, where the feedback is very specific because the system between the actions that it's able to take, the knowledge that it has access to, and of course, its ability to actually drive a plan that is really fine-tuned to the individual that needs the help. These things have come a long way to the point where I'm able to get the assistance I need, and then I can sort of focus my attention on the things that really matter for me.

Mark: 15:41

Okay, yeah, that makes a lot of sense. I think to add on to that, I was reading up about you before our interview, and you talk a lot about writing.

So writing, I found it on your blog, on your DevRev profile. You say that it's really helped. How does it help you in all of this and in clarifying the work that you do and giving feedback and. How do you use writing every day? And why is that important to you?

Ahmed: 16:07

Yeah, I know writing is important to me. And you know, I think if you ask any person who believes in writing and values writing, they'll say that they're not writing enough.

So I'll say the same about myself. But I think we do like writing things down. And I think writing has become more important than ever, because good writing and quality authentic writing is actually one of the big inputs to these new systems, because they are so natural language biased right now, that I think writing is more important than ever. My sense is, especially the way we've created our startup and a long And I would say like over the years, we've seen this actually with many of the other companies, even some of the larger companies trying to geographically get more diverse. There's a diversification happening. There's a globalization around staff, you know, where teams are not just in one location, one city, one state. Now you've got people who are living remotely. There's people who are being hired on in different sites. And that's just about accessing talent wherever it is. It's about really diversifying. Increasing your reach and understanding how to create better product. And, you know, that's those are the benefits of it. But the downside is that you need to understand how to offset time zones, you have to understand how to offset the fact that you may not be available and physically around each other all the time. And so that's where writing really comes in. I we really felt this even in my previous law, you know, job. Where during the pandemic, writing saved us from getting too fast and loose with our ideas. Frankly speaking, I think that people really start to understand that it's a great equalizer in terms of memorializing ideas and allowing people who aren't ready to sort of be vocal with their feedback to be able to provide feedback. I think these are really important steps.

Mark: 17:51

Okay. No, and I think you're right. I hadn't even thought about writing as a critical skill for writing your prompts right now, because I see a lot of those long prompts that are really specified really well, and some of us have shorter prompts. I think writing is a critical skill in the current day and age of AI, especially. I hadn't thought about that. It's an interesting viewpoint.

Ahmed: 18:13

Yeah, I think Andrej Karpathy, and I've actually shared this in a few discussions that I've had recently, you know, he said this in 2023, that the hottest new programming language is English. And, you know, the same can be said about different languages. I think the point, the sort of the higher order bit was that our natural language, our spoken language is now the way we are communicating with machines. And this is the big breakthrough, right?

So you go from 20 or 30 million engineers out there to understand how to use a machine. Who deeply understand sort of these esoteric machine languages like markup, markdown, C++, Python. And you know, that only scales with intention, right?

So you may go from 30 million to 35 to 38 million developers, as colleges and universities start to matriculate people who have gone through years of understanding how to write in these systems, be able to debug this And of course, if you are able to change and basically create a massive inversion, where now the machines understand English or Dutch or whatever language. Now, the natural spoken language becomes the input to the systems. Of course, you still have to understand how to direct, and you have to understand how to build consistency and how to get validation done. But it's all being done with a set of people who understand how to speak English or whatever the natural spoken language is. I think that's the major inversion, that's the major unlock where you go from 20-30 million to maybe 200-300 million developers. From my perspective, that's just one more reason why writing is going to become so much more prevalent and I think so much more useful.

Like being able to understand how to be intentional with your ability to direct a system, not just a human being, but an actual machine is going to become increasingly important. And being able to convey really good ideas and outline tasks and break down thoughts. And you don't kind of have some sort of written trail for writing. Meetings and transcripts, that just becomes more information for a system that can reconcile structured knowledge as well as it can reconcile unstructured knowledge. All of these inputs help it sort of refine its understanding of your business and really understanding how at the end of the day, customers, products and work are sort of getting interconnected. And the better it understands that the matter, the more useful it can be to sort of help you progress in your workspace.

Mark: 20:35

Okay, that makes sense. I think in the beginning you said you could be a software engineer if you were very intentional about learning a language, C++, Java, JavaScript. Would you say that people need to be more intentional then about learning their language, learning how to write in English or Dutch or whatever their language is? They should be more intentional about that skill?

Ahmed: 20:56

I think so. And what I would say, Mark, is that it's not just about writing as sort of a way of creating new workflows. Because remember, a lot of the writing that's going to be done is not to create new workflows, it's to actually create inputs to workflows that already exist.

So imagine in a, in an organization, or a department, specific department within an organization, you may have, you know, 20% of the time, you're going to be writing a new workflow. The people creating 80 or 90% of the workflows that are actually being administered. And that's, those are the types of ratios you typically see. You have certain people who are the creators and everybody else is consuming. And so just imagine that, you know, the rest of the people still have to do a lot of writing because they are still producing content that is going to go in and as either a feedback loop or as direct input into these workflows.

So I would say that as much as it is about understanding how to direct machines and GPUs to do things that closely match what your behaviors are that you're expecting or intending. I think it's about understanding how to rationalize and create good content that can serve as the inputs for better, more rational thinking happening by way of these automations.

So I think people have to think from both perspectives.

Mark: 22:07

Yeah, never thought about it that way. It's really cool to learn. Okay, we take it back to DevRev a little bit about the products. One of the things that I was really excited about and that I've learned from people like you all your presentations and DevRev isn't actually built on top of a vector database, or it's partially built on a vector database, but there's a knowledge graph underneath. I think a lot of people, this is a very tech audience, will understand the difference, but could you explain Quickly, the difference between those two and why that matters to DevRev.

Ahmed: 22:40

Yeah, I think probably the safest, simplest analogy you can make is that you can have an application built on top of a database. But until and unless you have the right relational mapping, the right schema, you don't really have value creation, right?

I mean, anybody can use a database, but depending on your schema, your index, your relational mappings, you're going to get. Different degrees of results. And so it's the same thing here. Anybody can create a vector database. In fact, we had internally in the early years before the vector database community had created really high quality products, we had our own vector database. And we actually worked with products out there and gave a lot of feedback so that we could get more optionality in the market. And now we actually use a pretty high quality product that took in a lot of feedback from us, which were completely happy to do. And I think the reality of the situation is that it's one thing to actually create a vector database, all it takes is using sort of a off the shelf embedding model, creating embeddings and putting it into the vector database as a series of mathematical vectors. But to actually understand how to, again, connect product to people to work, that's really how you create a business identity. And that business identity is going to be different for every organization. And how do you do that in a way that really kind of creates value for people and by way of helping them discover knowledge, take action on that, be able to exercise that knowledge? Express their ideas as insights, being able to express it as actions. That's really where the value is. And so I would say it's that schema, it's that relational graph that's created. You have to have that on top of the vector database. Very similar to what we've had to do in the past with business applications of a generation ago as well. And we do a lot there.

I mean, I think in terms of understanding how to connect those components, there's a lot within our system by way of just the way things are linked. It's about not only informational, like understanding the information as it's being used.

You know, sort of embedded, but understanding the relationship. So like, even if you have large documents, how do you sort of build relationships between the topic of the actual document versus paragraphs?

And then of course, how do you relate paragraphs with illustrations? And like, how do you take the information in that specific way? Document and connect it to a specific product feature capability? How do you connect it to a customer?

You know, just because you have an article doesn't mean it's connected directly to the customer. The customer may not even be mentioned, but you have to understand that there's certain semantics and there's certain relationships that are inherent to the document itself. And these are the types of things that I think we do particularly well that, you know, customers really take advantage of.

Mark: 25:22

Okay, and talking about those customers, you've said a lot about what DevRev is and that it's built, right? That it has all these different agents built in. But it's still a bit fuzzy what it actually does. Can you give me an example of what the coolest thing is that you've seen that one of your customers has built on top of DevRev?

Ahmed: 25:40

Yeah, absolutely. I mean, since it's a horizontal AI platform, typically, if you're going to sort of generate a knowledge graph, generally what people are doing is they're bringing, they're connecting different IT tools that they use today, whether it's their CRM or their work management tool, their customer support, customer experience information, or even sort of their suite of products that they use for managing meetings and calendars and any other things. They connect those things into our system. What I love is, you know, When people actually start to get all of their work observability automatically without actually breaking the flow of their staff.

So as an example, you've got engineering teams where the engineers, they don't even have to go and create issues anymore. They actually just continue to do work in their sort of development environment. And all they do is just say to create issue or update issues. Timeline and the updates go directly into the system and you can actually get those updates into the system and start to build automations on top of those updates so that all the observability makes its way into our product, and we can drive behaviors off of that.

So project managers, product managers, engineering leaders can actually take action on that information, even collaborate with people socially, without the person having to actually even use the product. I think that's pretty cool. We've had a lot of people that have been using our product to actually do ideation, where they'll say, "Hey, look, go do some research on a specific..." In fact, I did a live stream on this some weeks ago, where we had a junior PM that basically put in a couple of sentences and into the chat and said, "Hey, listen, go do some research into something that we should be working on right now." And the system literally went in and scavenged the opportunities, the outstanding tickets, even meeting transcripts, and actually came back and made a series of proposals. Person actually selected one of them and said, "Hey, I like that idea. Can you actually create an entire initiative out of that?" And that gets created and then you just say, hey, listen, just go create that as an enhancement record in the product. And now all of this ideation that's been happening offline kind of gets codified as a record in the product. It's connected to the right feature. It's assigned to the right person. It's got the right stage. And now you can actually start an internal dialogue. And in fact, you can have an automation that can actually break down that idea into the series of issues that can get now triaged. By the right teams. And so everything's getting triggered by two, three, maybe four sentences from this product manager. And those are the kind of experiences that I find remarkable. But of course, just in terms of business value, not thinking bottoms up, but maybe even thinking top down, there's a lot of businesses that come in and say, "Hey, listen, why are we even using product X or Y for CRM? Why are we using product X or Y for support?" why don't we just start using this product here? I think it makes us really proud when customers come in and start to use our product as an enterprise search tool or as a business intelligence, team intelligence tool, something that can allow them to take action on the information, the insights that they're getting through our knowledge graph. But then quickly they start to say, you know what? You've got the record view, you've got the right surfaces, you can meet us when it comes to customer emails, customers in Slack, customers. Over telephony, customers, anywhere, you can do all of that. You've got the right surfaces, and you've got the right product in terms of record views, analytics, discoverability of content, and sort of being able to codify system of record information and even unstructured information like knowledge base and articles. Why don't we just use your product for a very specific vertical application? And that's usually a very interesting proposition. And that happens quite often. And I suspect over the next three to five years, there's going to be a lot more of that as well.

Mark: 29:28

That's interesting. I like the part, and this was way at the beginning, where you said that engineers no longer have to go into their ticketing system to update their tickets. And I think so many engineers are going to love that. I think this is probably the systems and the processes that they hate most.

So I think they must all be very happy with that. But that technology no longer having to do.

Ahmed: 29:47

That. They are. Mark, and I'll tell you the message that I'm often harping within the company is that work observability, meaning the ability to observe what's happening on the work front should never get in the way of the work itself. And if you think about any department, if you're talking about a field rep believes that field rep is serving the company by being out there in the field, by meeting customers and prospects for lunch, maybe dinner, by holding a fireside chats, by really getting in and understanding what their needs are, why they're here. They're not happy or why they're delighted with the product, whatever the case might be.

So field reps are trying to upsell, cross-sell, and sell to new buyers. That's how they serve the business. They're not serving the business by actually stopping their car on the middle of the road and trying to upload a bunch of information, right?

So if that information to a higher degree can get into the system authentically without them having to constantly sort of create those stoppages in their work stream, it really makes them feel a little bit better. You know, they go from meeting to meeting with a lot of energy and, you know, they're in that state of flow. And I have to tell you, as an engineer, I know that firsthand, like when you're in the middle of trying to solve a problem, the last thing you want to do is actually minimize the screen, go into another screen and start going from thinking about writing software to writing and software, an update about the software you've been writing. And it's, you know, And I think that those are the types of things that are going to be major upgrades to people's processes. I often talk about how in this new world, we're going to start to reduce the gap between human observability and machine observability.

Like Mark, you know this, there's a great degree of machine observability that's really come together over the last 15 years. So if you think about logs, you know, the type of information that goes to Datadog, or the type of information that might go to Mixpanel, or Amplitude, or any of these tools where, you know, usage statistics, or login statistics, or error statistics, that information, we've been collecting this for an entire generation. And we never thought about it. And we never said, Hey, why doesn't a human actually stand there, watch what's happening in the system, and actually, you know, manually insert that. Metric or that measure. We would never do that because we know that actions that are happening online, they're happening at a scale where it has to be published automatically. It has to be published automatically.

So a microservice publishes its logs automatically. No human is in the way.

You know, an action that happens on a web app or on a mobile app, if there's a completion of a there's gonna be a metric emitted that says, this was a checkout that got completed for a net total of this much money, and it took this long, or and this was the number of things that were in the shopping bag. This is the number of things that were removed from the shopping bag. All of that information gets collected automatically. And that's what I would call machine level observability. And what's happening is that we're able to expand that machine level observability into more and more human tasks now.

So just like, a person who's on a website that's buying something doesn't have to end the session by actually providing a summary of what they bought and what their experience looked like. That would seem inane at this point. It's the same thing here and now with developers, field reps, customer support agents. Increasingly, they no longer have to summarize what they did. Those actions can be collected automatically, and those behaviors can be understood automatically, and those metrics will be collected, I would say, in an increasingly progressive way And that's just going to make it more interesting and more joyful for us to just do the work that we do. And I think these are the types of things that I'm particularly interested in.

Mark: 33:35

Yeah, and I think that's going to make a lot of people very happy. I heard that story about the field trips earlier from someone else as well, and they're going to be so happy not having to update their CRM every day or every Friday afternoon or whenever they do it. And I think the other part of this is that all this machine observability, we used to have these enormous log databases that nobody knew what to do with, just because there's too much data in them. And I think their AI is really going to make a huge difference as to how we can analyze all those logs as well.

So it's going to be an interesting time for sure.

Ahmed: 34:10

I'll tell you one other, like before we switch topics, I'll say like, you know, people talk about sort of the bias and, you know, the risk of hallucinations when it comes to large language models. And, you know, we take that very seriously here at DevRev as well. And there's a lot of work to be done in the years to follow on that front. But I'll tell you this much that, you know, humans bring their own cognitive bias. And one of the ways you see that in a very way that's quite jarring is in the quality of the type of updates you get from humans.

You know, even if they're you know, two engineers that are writing software, they may be submitting software into the same exact repository. However, their updates, just the templates they follow, the categories of work that they describe, the type of assessment that they may have can vary. And what that does is it creates unnecessary bias into the system. And those are inputs that are going to go into actions, insights that more and more automations are going to take on our behalf. And the more we can sort of create more consistency on the work by creating, By having automations that are effectively creating those updates, I think it starts to help us on sort of subsequent tasks or the follow on actions on that information.

So that's going to be really important. And what that does is it allows us to actually take our personal styles and our distinct tastes and apply it in ways in areas where it matters more. Right.

So I think that's going to help us differentiate in ways that are actually positive, not negative.

Mark: 35:38

Not more admin work to do for all of us, thankfully. Absolutely. Okay, now indeed I was going to switch topics. I really want to talk more about your development team. You said you have a development team of 100 plus engineers. They're spread all across the world. You said a little bit about how you manage that, but I think I'm a big fan of remote work. How do you coordinate that? Because that must be... Quite different from having them all in the same building, which would probably be very nice. How do you manage them?

Ahmed: 36:07

Yeah, it is nice to have everybody in the same building. And sometimes we have most people in the same building at a given time. But it's a remote team. And you know, you said 100 plus is actually 200 plus.

So we've got a number of people here. And I have to tell you, one of the things that we have to rely on is our product.

So DevRev uses DevRev in a meaningful way. So search analytics and workflows, they matter to us, you know, in every facet of the word. I often use automations to do theme level aggregation.

So there's a tremendous amount of work that goes into us understanding, are we reaching our goals? So you have to be able to interpret what a goal was. If the goal is not clear, somebody has to take the action of clarifying the goals. Once the goals are clear, we try to understand who are the people who are dedicated to reaching those goals? Who are the shared resources on those goals? And we sort of thematically understand how to cluster people towards these themes, these rocks, these pebbles, the you know, the sand work. And by doing so, we understand and how to sort of connect people to real outcomes over periods of time, right? Over a month, over a quarter. And by doing so, we sort of make sure that there's enough direction, there's enough oversight that's coming through, and things can get escalated. Most of the time I can actually review the themes. There's typically not that many themes. I can actually even connect with people on a biweekly basis across all themes.

And then between that and making sure that people are doing monthly retrospectives within their pods. And then of course, at the quarterly level, as we come together with the core leads and do planning so that the next level of themes, the rocks and the pebbles are planned together, and those outcomes are clarified to some degree, and then you get to 100% over your own next several weeks. It seems to be a pretty good process for us. We do still appreciate that face-to-face contact.

So we try to make sure that folks like myself and others are visiting. Different sites.

So I'll go to Argentina, I'll go to Europe, I'll go to India, I'll go elsewhere. And, you know, I'll meet people and, you know, sit with them and just make sure that they understand what I'm about. And I understand what they're about. And I think that helps, you know, sort of make sure that human connection is not lost. But, you know, that human connection is no longer about, you know, me asking for, you know, dubious updates on a weekly level. It's now about something that's more important to the business. It's about, you know, understanding how to do outcome-based planning. It's about understanding how to create ambition. It's about understanding how to sort of unblock ourselves by creating a better toolkit for ourselves and, you know, creating playbooks for consistent behaviors. And those are things that I feel I can do a better job of conveying. I can write some of those things down. I can help sort of talk about them when I'm there in person, understanding how to create influence, understanding how to help each other, understanding how to build on each other's ideas. I feel like I'm being, you know, more valuable to the company by focusing on those things. And that's kind of what we do.

Mark: 39:11

So I'm on the human aspect on the actual admin. Updates, retrieval.

Yeah. Yeah. I understand.

Yeah, exactly. You have this really large team of engineers, 200 plus, as you said, You've grown into leadership over the years. What are your core beliefs? What kind of a leader are you?

Ahmed: 39:30

Yeah, I think I try to bring some kind of imaginative intelligence. I am generally the type of person that sort of tries to, you know, have a person think about how to go from a three star experience to a five star experience. And I think for me, that's my way of thinking. Teaching is by helping people understand how to develop ambition. And I try to lead with positive energy and enthusiasm for the work. I think that works best at companies like DevRev, startups, where it's all about where we're going rather than what we've done over the last 10 years or 20 years.

So I think it's important for people to align what they're naturally doing. Good at or what they naturally bias towards focusing on with what the needs of an enterprise are. And for me, you know, my bias towards thinking ahead and understanding how to do things really well, I think aligns with what DevRev has to do, which is think big and make sure that whatever it's doing in a way that it has a certain level of refinement and sophistication so that we can continue to meet the needs of what customers are expecting from us.

Mark: 40:41

Okay, so cool. Yeah, I think it's very different from working for a corporate or somewhere else, Anita. A scale-up like DevRev is very different to work for, I'm sure.

Ahmed: 40:52

It is. It is. And you know, I think you can always be yourself in any enterprise. But of course, what aspects of your skills, your talent, your toolkit are being exposed and leveraged, make, you know, defers. And, I have to say, you know, obviously, having worked at a much bigger company prior to this, I was still able to be myself and sort of apply myself to some extent, right. But, you know, when you have a company like this, and you're able to kind of have that much more command over what's happening, and that much more influence, it just makes it more satisfying, because you know that, you know, whatever is working, you had a hand to play in that. And, you know, whatever is not working is your responsibility to correct.

So from that perspective, you know, being hands on matters, I think, has been a big part of my career. Having some level of optimism for the future has to be the case. And that's kind of where we're at.

Mark: 41:41

Okay. Bon. Okay, I think then from the team side, taking it back a bit, I want to wrap up with a couple of AI questions. Sure, of course, because that's, I mean, it's a topic of at least what DevRev was built on, but what everyone else is talking about today.

So you guys are completely built. AI first. And of course, you've DevRev for a lot of your development processes and the observability that you talked about. But how do you guys actually, guys and girls, all the engineers do their coding to use AI tools as well? What's your experience been with those tools and where do you use them?

Ahmed: 42:15

Yeah, absolutely. And it's not just the engineers.

So we approve, we automatically approve. All cursor licenses for engineers. Program managers, PMs, designers, and several other functions within the organization. One quote from Scott Belsky from a couple of years ago that really resonated with us is that AI will collapse the talent Slack. That's what we're seeing. We're seeing designers increasingly interested in writing code because you can Sort of vibe code your way to a functional prototype if you're a designer or you're a PM. And we're seeing a lot more of that right now.

So we're seeing a lot more functional prototypes that people are using to understand how they want to build an experience. You're seeing a lot of like websites being brought up to actually. Give even customers or prospects the opportunity to see, okay, what features will they use? How much of the feature will they use?

So you'll see these sort of websites that we bring up for our customers where they can use these slider tools to figure out, okay, how many customers do they anticipate? How many requests of this type? Or how many tools do they anticipate?

So really clever things that even some of our solutions engineers, some of our pre-sales folks are using to very quickly stand up powerful experiences And I think they run the gamut truly. And of course, for the core engineering, it's really important to have the best of the best tooling. What I love about these tools is not just the fact that you can ask questions like, hey, what is this piece of software doing? Or I want to write a function that does this, that, or the other. But it's about understanding how to have just the best of breed experience.

Like now in these modern tools, you can actually build get debugging help, you can actually have an inbuilt terminal. There's a lot of like little experiences within the modern sort of, you know, sort of these forks of VS code that have come out.

I mean, we use cursor, but there's others out there. And they've just become increasingly powerful.

So yes, we use a lot of these tools. But we use other tools like Replit and lovable and many others that are pretty AI forward.

So in general, I would say yes, we're all in on these tools.

Mark: 44:27

Okay, that's really happened fast over the last one or two years, I would say. I think when you started with DevRev four years ago, there was no AI software engineering, it didn't exist.

Ahmed: 44:39

No, absolutely not. I mean, in fact, it would, A lot of times people will write about how it's about maintaining the code and not about this and that. But I have to tell you, code comes in many shapes and sizes. And I think to allow people to... To have access to be able to build code that just wants to serve a very specific purpose. Because the time value ratios have changed now, because you can just describe what you want to stand up. You might have code for things that you would have otherwise never written code for. And so, and this is not code that you want to maintain for six, nine, 12 months. It's sort of the fast fashion equivalent of code, right? It's not code that's actually serving business application function. It's actually doing something very different. And so these are some things that we have to understand about where we're headed as a society is that we're going to start to write code for things that We couldn't afford to write code for before. And it does not have to be code that needs to last for three years, five years, seven years. That used to be true when you had an engineer that had been trained for five years or seven years, and you had to make sure that whatever they're doing was sufficiently complex and was going to be for the purposes of building a very specific microservice that served a very specific application need. It's not like that anymore. There's a lot of software being written for many purposes within our enterprise. And I think that's a good thing. And of course, even the software that has to be maintainable and available, you'll see it's surprisingly easy now to build up a unit test coverage plan for systems.

So we're seeing that ever since some of these tools became very prevalent, we saw even in areas where our coverage was not where we wanted it to be, it was rather easy for us to get it up because the system understands, the software understands which paths are not being tested well. And I'm not saying and it gets you The absolute best test coverage plan. You obviously don't want to rely on it completely, but between good intention and just having better tools, that combination can get you pretty far.

Mark: 46:39

That's yeah, I understand. Quite a bit of those smaller things myself that i've happily thrown away again so yeah thanks so and i think this is where we are right now you're a parent i read of two kids I mean, there's a lot of future ahead of us. The past two, three, four years in this space have been absolutely incredible. I think the future is very hard to predict, but if you had to make a guess, what would you say about the future? What do you tell them about the future of AI and where everything's going?

Ahmed: 47:12

You know, it's amazing. So my son's going to college next year and my daughter's going to be a junior in high school.

So... They know a lot about AI. And in fact, I had this sort of humorous conversation with my daughter.

You know, she had taken computer science last year, and I was just asking her, you know, about AI. You know, computer science. And she said, you know, I don't think we're going to be writing code like you did when I, you know, when I'm sort of, you know, past college. And she's probably right.

So I think a lot of it is not about what I'm telling her. It's about what she's telling me. And it's amazing how it's really changed everything, even at the high school and college level where people are not writing the way they used to. They're not brainstorming the way they used to.

d it was because my son has a: 1999

So that tells you a lot about where it is in terms of just popular culture and understanding where it is and as an educational tool, but as an entertainment tool. So things are changing very quickly. I think, you know, the youth understand and they've completely embraced these technologies in my mind. They've embraced them as part of their academic, their business, their professional lives. And of course, I think as it pertains to their personal lives, it's going to become increasingly valuable, especially as these things become better from a privacy standpoint and start to make their way to the edge. And as models get more and more personalized and we're able to undertake more long horizon tasks with greater independence, and those things are happening, there's a lot of research in terms of the reasoning and planning and memory and all that.

So as these things get more sophisticated in the years to follow, I think it's just a matter of. Bye time.

Mark: 49:22

It'll be very interesting to watch and cool to watch for us and for your kids for sure as well. Absolutely. I think with that, I'd like to wrap it up. Before we go, where can people find more about you or about DevRev? Where should they go?

Ahmed: 49:38

Yeah, absolutely. They can join our Discord server if they really want to get in on the action. Otherwise, we've got our DevRev YouTube channel.

So they can search for DevRev on YouTube. They can search for DevRev on the internet at devrev.ai. And of course, you know, they can find me on LinkedIn, my name. You can provide the LinkedIn link, I'm sure. And yeah, I'd love to connect with people. And yeah, DevRev is everywhere.

I mean, it's an international company. And I would say just search for DevRev and stay connected to us. And hopefully we have something for you.

Mark: 50:07

Okay thank you very much it was very nice having you we'll provide all the links in the show notes for sure and yes thank you lamed very nice having you As we wrap up another episode of the CTO Compass, thank you for taking the time to invest in you.

Ahmed: 50:16

Thanks Mark. This was a pleasure. Appreciate it.