Ep 51: From Automation to Agentic AI with Nextdoor CIO Shyam Bhojwani

Evan: Hi there, and welcome to Enterprise AI Innovators, a show where top technology executives share how AI is transforming the enterprise. In each episode, guests uncover the real-world applications of AI, from improving products and optimizing operations to redefining the customer experience.  I'm Evan Reiser, the founder and CEO of Abnormal AI.

Saam: And I'm Saam Motamedi, a general partner at Greylock Partners.

Evan: Today on the show, we’re bringing you a conversation with Sham Bhojwani, Chief Information Officer at Nextdoor.

With millions of users across the U.S., U.K., and Canada, Nextdoor is one of the largest neighborhood-based platforms in the world—connecting communities with real-time information, local recommendations, and public safety alerts.

In this conversation, Sham shares his perspective on how AI is shifting automation from static workflows to dynamic, context-aware agents, why IT and cybersecurity are prime candidates for fast AI wins, and finally how culture, architecture, and data readiness all shape the long-term success of enterprise transformation.

Evan: Yeah, maybe to start, do you mind giving our audience a little background about kind you and your role today?

Shyam: Absolutely. Thank you for having me here, Shyam Bhojwani. I am new to Nextdoor. This is month number three. I have the title of CIO and head of security. So I get to do both. Uh, I've been fortunate to start my career as a developer. So being a developer, then getting a chance to work on Wall Street, sort of running business systems, business technology team. So from...

Like I used to be a developer at KPMG, not a career that a lot of people start with, but that was my journey, right? Like writing C-sharp .net code and sort of transitioning from there.  Uh, Then working at Wall Street, like doing everything IT, right? Building automations, back office, onboarding, offboarding. And that actually got me a job at Peloton, the, the famous fitness company during COVID time. So I was there for almost like three years. And after that, I worked at a company called Vocado, automation and integration platform. Uh, So that gave me a breadth of sort of exposure to different things. 

So typical dev to IT and security transition, like that's been my career so far. Again, as I said, new to Nextoor. Still drinking from the fire hose.

Evan: It's probably one of the more familiar brands, but for any people in the audience that haven't heard Nextdoor, can you share a little bit about what do you guys do?

‍Shyam: So Nextdoor is a neighborhood platform. The idea is we want our neighbors to feel connected, have this community aspect. So we are a community driven company. Mission is to keep neighbors connected. Uh. Whether it's a fire alert in your neighborhood or whether it's about a soccer game happening in your neighborhood, people post things on Nextdoor, right? Staying connected. So that's like the user aspect and our sort of business model is around ads.

‍Evan: Can you give our audience like a sense of scale, right? Cause I don't think people appreciate kind of like the scale that Nextdoor operates at.

‍Shyam: Absolutely. I think we, as I mentioned, we are a community company. So again, not jumping to like direct numbers, but millions of users on the platform, right? Using the mobile app, whether it's your Android, iOS. So I would say very popular across US, UK, uh, Canada. So these are the markets primarily, but millions of users, lot of posts, lot of messages that every interaction leads to. So it's a big scale, heavy processes, heavy integrations to think about, but large scale.

‍Evan: So one of the reasons I was really excited to have you on our show is because it was because part of your background at Workato, right? And so we'll talk about kind of how AI can transform product and customer experiences. But I think for a lot of CIOs, what's top of mind is, how do we transform some of our business processes? And when I say transform, I don't mean digitized. We've gone through a decade of digitization, kind of taking an existing process and making it better using software. But doesn't like with AI, there's now an opportunity to actually transform, rethink some of these processes. 

What do you see as the opportunity there, right? And like, what do you think maybe most people are underestimating?

Shyam: Decade ago, IT used to be the name of department and it moved to business technology right like being close to business so, uh, think about digital transformation. We all have seen like transformation around. You had your typical processes. Then we had this, uh, integration journey, automation journey, right? Like that was still 2023. 

Now with AI, all things are sort of coupled together in my opinion, right? Uh, AI is continuing to be the user interface, but now with like AI automation, it's a tighter, uh, I would say combination together. And in simpler terms, like I'm like super bullish. You can see the excitement on my face, right? Like working at uh, Workato seeing the journey. 

This whole agents, they are powerful, context-driven, and with context, they are able to orchestrate the right backend processes, right? Like earlier, you would sort of create static processes that step A, step B, step C. Now with AI, it can go from step A to step C based on context, right? So this static, automation to dynamic automation. I think that's where AI is becoming more and more powerful, right? 

So all to say the shift from your typical automation integration to agentic orchestration. And that's a fancy word everyone's using right nowadays, But I think that's what the future is. Your agents, whether it's your front desk agent, whether it's your IT agent, your security agent, your procurement agent, right? All these agents are now able to talk to each other with the right context. I think that's where the real power is. But goes back to the discussion that data is important. Without data, these agents are not that powerful.

‍Evan: I've been working in machine learning AI for like 15 years and, um, I find myself continually surprised by kind of how fast things are going. Even, you know, for our own kind of AI transformation, six months ago, we're like, get everyone on GitHub copilot. It's ridiculous you’re not using copilot. And then like in January, we're like, copilot's obsolete, right? We should be using cursor agentic, you know, stuff. And now like this month, it's like, wait, forget using cursor, right? Why do we even look at like source code, right? We should be, you know, we should be even higher up in the abstraction. 

So I find myself even as like, so like, it's probably more knowledge with the average person about AI applications. And obviously we build AI applications, I keep underestimating what the real opportunity is. 

You know, what do you like, I guess there's, there has to be a lot of CIOs out there that probably are also underestimating like the real opportunity. are, what are maybe some, are they kind of concrete examples where you'd say, Hey, most people think probably this is not possible, but actually like, you know, given a little bit of time, we're going to get further and here's what the real opportunity is for, you know, I don't know, some key pieces of, of, you know, business process transformation.

‍Shyam: No, no, that's an excellent question because to my earlier comment with AI everyday, it's a new learning, but also a lot of unlearning, right? Like what we thought was not possible. Now it is possible as of today, right? So I am with you on that every day I wake up, I hear about all these, like, the cursors, like new products, new agents doing so many powerful things. 

Uh, The way I am sort of thinking about this again, fortunate to get that. automation integration experience from my past role. The sort of thought process there is with AI, you can think about all your processes, right? And don't just think from point to point, task to task, think big picture, think, uh, joining the dots sort of mentality, right? 

Like for example, I'll give a very simple example, right? Your typical IT uh uh process where you have an employee they start on day one, they should get right access on day one, but as they spend time like day one, day seven, day 30, they would get elevated access to a lot of these systems, right? So rather than thinking about this that, I'm gonna get them access the right way on day one, there needs to be constant learning. There needs to be a context that this person may be part of a developer team and they need to get, uh, their behavior should be learned throughout logs, different telemetry, and AI agent should automatically figure out that, now you have completed 30 days, maybe you get access to, uh maybe some specific tool, right, cloud system. So that context-driven business process, AI orchestration is like one way to think about. 

And I'll also say this, there's no right or wrong answer. You have to start somewhere, but rather than having like a task by task point to point sort of automation AI journey, think big picture. I always give this analogy to my team, right? Think about different objects, think about graphs. That's how your AI agent should interact with each other. Now AI agents are your GraphQL objects talking to each other.

‍Evan: From your experience, what's your list of like, Hey, like if you're looking for a couple of easy wins, right? You know, start with A, B and C, right? Is it, don't know, customer support? Is it knowledge-based generation? Is it, you know, um, making it so anyone can update the website right under AI, under an AI coach. Like what would be your kind of like top three, like if you're looking for a couple of like fast, easy wins, know, where are some baby places to start?

‍Shyam: My first sort of recommendation is IT is a low hanging fruit, right? Like your typical help desk, getting right access, having the right context that this employee belongs to a particular cost center uh and getting them the right access, right? Right access, right approval. Everything should be real time. I think that's like an easy win, right? Like easy win, easy to sort of implement in my opinion, right? 

So. Same thing with cybersecurity in my opinion from a cybersecurity standpoint, you get so many alerts. You want those alerts to be to be sort of handled sort of from a source standpoint, right? Like your investigation, your triaging and sort of remediation. I think these these two are like your first easy vents from from like a back office standpoint. 

And then I would say your typical, to your point, right? Customer calls. My favorite example that I always talk about is, There are so many meetings throughout the day, whether it's for customer facing teams or non-customer facing teams, right? And for customer facing teams, having AI to sort of do transcription and create meeting notes. And the main thing is if a competitor's name is mentioned on a call, automatically displaying like a battle card. Like that's such an easy win, right? This name was mentioned, just go to your notes, right? And it helps revenue. Easy win, easy to sort of convince your sales ops, rev ops team, right? So I would say that would be like a second easy win. 

And third thing is, uh, I don't use the word deflection. think deflection is sort of negative in my opinion. It's about delivering that experience, right? Whether it's customer experience. I don't know, like, have you heard this phrase XLA, experience level agreement? A lot of my peers are talking about XLA, right? Like with AI, don't just focus on result, focus on experiences. You want to give a good experience to your whatever, whether it's customer or employee, right? So use cases around that.

Evan: I kinda wanna go back to this example you shared with like kind of the, you know, listen to the call, show the real time battle card, right. Not, not that specific example, but like that's a really good one where, sorry, it's a good example I think of more like transformation versus digitization, right? If, you know, if every sales rep had 10 interns where they're following around all day. Right? would have all the materials ready to go, ready to put in front of the, you know, in front of the AE or the salesperson, but like you couldn't, it was hard to kind of actually digitize that in the past. Um, but now kind of this idea of like those 10 interns that could help out that person. Well, now those could be like, you know, your one to 10 AI agents and there's now things you can do that you couldn't do in the past. And I think that, that, that example is also good because you're not actually really disrupting or changing the current process, you're actually just like augmenting in a way where it is kind of transforming, but doesn't require like a whole process and change management, right? 

Are there kind of other examples you have in that bucket? I just think that was such a great one that certainly, that should belong to everyone's like easy win bucket.

‍Shyam: Another example that's like very close to me is, uh, sort of having a sales AI agent. So for example, along same lines, right? Salespeople, they have their pipelines, quota, revenue, busy job, right? And if they have to go from their CRM to their other systems, right? It's a lot of context switching for them. 

So instead of like, for example, if they want to figure out what's in the pipeline, And like simple example, right? I have a call scheduled in 30 minutes with the prospect, right? I'm gonna go through my notes, prepare, but if the AI agent can automatically look at my calendar, be like, oh, you are meeting with this prospect customer in 30 minutes, it automatically goes to your CRM, gets the internal context, right? What is the internal context? What discussions were done? And also goes externally to figure out was there something new about this prospect on on on the internet? Are they making any big announcements? So sort of giving that internal external context and right before the meeting, preparing you in less than two or three minutes, like how cool will that be, right? We all have so many meetings, we are jumping back to back, but if the AI agent can have the internal external context, and get you up to speed, that's a big productivity gain. So that's my other favorite example that I talk about.

‍Evan: I think another challenge potentially is the cultural challenge, right? People need to learn that, or people need to like imagine a world where AI can be pushing information, you can pull information. And I think if you and I had, if we did the eight hour podcast and we envisioned what does a company look like in 2030? How does it work? You know, there would be some big changes, right? But I think one of the barriers to kind of evolve there is the kind of cultural nature, right? You got to kind of introduce this, right? 

We even seen this in the software engineering team, right? There's tools that are so obviously good, right? Cursor, Claude code. But sometimes you have people are like, no, like I kind of like doing things the old way. So there's going to be some version of that and kind of any like piece of the business you're trying to transform. What advice do you have for people that are, about how to kind of get the team ready or kind of introduce things that are, that kind of show value without too much disruption. So people start pulling more of this from kind of technology teams rather than feeling that's kind of be forced upon them.

‍Shyam: So the way I'm sort of thinking about this is first thing to sort of clarify is AI is not here to take anyone's job. AIs actually help here to sort of amplify what you do, give you time back so that you can focus on bigger things, complex things, right? I think it starts from that. Like it's more of a culture thing that everyone says that we are a fast-paced company, we are a tech forward company, but now it should be, we are an AI forward company too, right? So I think it starts from that sort of culture in my opinion.

And the second thing is I'm a big believer in having office hours across your company, right? Again, wearing my typical IT security hat, office hours are a great place to showcase people what's possible with AI, right? Like, just like simple thing, your company has 10 departments, start with two departments, pick two use cases, be like, oh, this is a tool we have, this is what you can do from a productivity gain standpoint, right? So I think your employees, your audience appreciate that like their day to day processes have been talked about and AI is able to augment what they do and sort of give that productivity. 

So starting from there, big believer in having newsletters, sending out Slack communication regularly, right? Because if you don't talk about what's possible, your employees, your customers may not know about it, right? So there needs to be a big emphasis on enable man communication and getting people excited, right? Like, 

I would say it also starts all the way from the top. Like if your company has like a OKR starts from there, right? Like, to my point, we are not just a tech forward company. We are, we are an AI forward company too. So it starts from there. I mean, I can talk about it throughout the day, but these would be like three or four high level strategies that I have sort of, applied at previous companies, work in progress at my current company, right? But if you position AI, automation, digital transformation the right way and sort of get it into practice along baby steps, right? Everything needs to be planned, strategized, right? I think it has a huge potential.

‍Evan: I want to go back to something you said earlier, if you want to be good at AI, you got to get good at data, right? You got to have it logged and organized. So this is kind of both a data architecture and a bit of a cybersecurity question, right? We're going to this world where, you know, hopefully both of our economies, we're gonna have all these agents doing all the mundane parts of everyone's job. They're going be very data hungry and they're not only to access, you know, they need access to stuff and you don't want these things to go crazy and get out of hand. 

So how do you think about like from a, I guess, like a data architect or technology architecture, presumably there's some centralized place under the CIO where data is going to be organized, managed and accessible. You're going to appropriately control access to different internal AI applications into that. How do you think about setting that up?  And what are any principles, or as you're thinking about how to assemble that for the future of these AI empowered workflows, Like, like how do you, how do we all make sure we're not going down the wrong direction, right? When we kind of set up all, you know, everyone's kind of do, we're all kind of working on the same thing, right? How do we make sure if we don't make mistakes or make sure we're all kind of going in the right direction there.

‍Shyam: I always refer to this triangle. You used to have people, process, tech, but I think now with AI, it's people, process, tech and data. Like these three are four important sort of pillars in my opinion, right? And specific to your question, I feel the main thing to sort of plan for is your data hygiene. 

If your data is not properly structured, if your data is not properly organized, AI is not gonna be that effective, right? So going back to the basics, I always give this example, your systems like your CRM systems, they are touched by so many different teams, right? And if they are touched by so many different teams, your data may not be at like the hundred person throughput, right? Like there might be missing dots there. 

So what, the way I'm thinking about this is data lineage to think about, right? Like for these AI agents, uh, start with a small subset of data because that's like a true litmus test for you. Because if an AI agent with that small uh uh data sample, if it's not able to do the right thing, that means your data needs to be rethemed, rearchitected, or maybe repermissioned from a security standpoint too, right? 

Like AI agents, it's good what they are doing, but if they are being over-permissive, that is a risk to your company too overall, right? So that's what I always say, you need to think about different data aspects, like what's the source? And when AI agents read, at which layer are they reading from? 

Like there's this famous data architecture, uh brown, silver, golden layer from a data perspective, right? I think for AI also, there needs to be like different layers that it can read from. And if it makes a bad decision, you should be able to reverse engineer using data lineage. So like typical data governance, data architecture, right? But having checks and balances in place, right? Like every time it makes a decision, being able to reference back, like how did it come to that conclusion, which layer is it reading from?

And now the interesting thing is gone are the days of thinking about like data warehouse, data lake. Now it's chat, it's unstructured data, it's everything, right? Your SaaS. So it's a bigger surface area to think about, but having these simple checks and balances is super important. 

And you actually remind me of something. Another sort of, I've been talking to a lot of my peers. You know what, many of them are doing? They do a end of day AI agent validation, meaning whatever the AI agents gave as an output, they compare it against different models just to make sure that if this was valid. So that's like a good validation, and it gives you like, is the AI agent making the right recommendation? Is the validation score improving on a day-to-day basis? Right? So baby steps, but standard data architecture to think about.

‍Evan: You're have all these agents, right? Doing different things. Some of them would be very, you know, minor tasks. They say, they're talking about the meeting prep bot, right? Getting a report at the end of the day and you're kind of agent, daily agent checking and saying, how many meeting reports did you generate, Right? If that number's a million, right? Well, maybe it's bad. it's zero, it's probably bad. If it's in some zone, it's, you know, it's not totally messed up. So there's probably some basic things like that, when you think about kind of the, you know, I know if you want to call it like the governance or the management of like your digital workforce, right? That is like a simple thing. Just like, hey, at the end of the day, send a little, you know, submit into this spreadsheet, like a little summary, that kind of characterize the goodness or the just like a litmus test of like, it even doing the right stuff? Is it even working?

‍Shyam: I mean, AI is powerful, checks and balances are still needed. Your validations are gonna make it better, so 100%.

‍Evan: I think a lot of companies, including ours, are struggling on organizationally, like so I think we're, I think you and I are likely, I'm reading between thelines, you and I are lying to like, the future is gonna be more AI empowered when it happens, how much it happens, to be determined. But I think a lot of companies are struggling to, lot of CIOs and CEOs are struggling, hey, how do I go activate that transformation, right? How much is centralized under IT? How much is kind of federated out to different kind of functions and departments? 

You know, what do you think the right organizational model is, right? Certainly for some things that you described about the kind of the data management, the governments, the security, that's gonna be more centralized as appropriate. You know, for, don't know, the sales meeting bot, right? Do you have a AI engineering pod under sales? Do you have it under IT? Do you have a business analyst that reports into sales that kind of works with the AI product manager? Like, you know, how do, how do you think about kind of the organizational structure that's going to most effectively kind of drive that transformation?

Shyam: One thing that we are sort of considering internal is there needs to be an AI ops team within business technology or IT, right? So the job of that team is to sort of talk to all your stakeholders, whether it's your HR,

finance, your sales, rev ops, your customer support team, right? And like the main goal is two things, right? Number one, you want to not have duplicate solutions across the company. And that's happening across the industry, right? Like your specific departments are buying AI products. There is similar product being used by another department, right? So number one thing is to sort of avoid the obligation across different department. So that's why having this centralized sort of AI ops team is a good starting point in my opinion. 

And the second thing is when you think about these different LLMs AI agents, output from one team or one department could be input for another team or another department, right? And there needs to be a team was able to join those dots. That's where this centralized model comes handy, right? So I think the phase we are in, the journey we are in, it needs to be centralized to start with in my opinion. And as teams grow as this sort of exposure to your different departments, business team, there could be like a dedicated point of DRI on those departments and team, right? But I'm a big believer this AI ops concept centralized team to start with because AI is moving at a lightning speed and you need like a catch-all team to start with, someone who can avoid duplication, better structure, better investments, right? 

And it's the same AI journey, right? Sorry, automation journey that we saw with integration platforms. First IT or business technology used to build it. Then you had like your builders within your business teams, right? I feel it's gonna be the same journey in the long run.

‍Evan: Are there any specific AI use cases at Nextdoor that you're excited about, right? Either kind of past or kind of upcoming future, whether it's kind of in the product or whether it's in kind of operations like, or anything you've seen, right? As you kind of, you're coming into the organization where you're like, wow, that's pretty bad ass, right? This is cool. Like we're already going in the right direction. Like, yeah, maybe brag a little bit about kind of cool stuff you guys are doing.

‍Shyam: For us, AI is in two buckets, right? The first is the AI within product. So we have AI for uh content generation. So if someone is doing a pose, AI is there to sort of guide them with generating uh content, right? So a pretty standard, simple use case. But one use case that I feel I am excited about is the one that I spoke about, your sales AI agent, right? We are actually sort of trying it out. We all are excited that it's gonna help our sales revenue team because they can just interact with one AI agent. That AI agent has complete context. I'll sort of talk more about it, right?

Let's say the person who's interacting with this AI agent belongs to the West Coast, right? The AI agent already knows that there's a territory-based division between West Coast, East Coast, for example, right? So the AI agent, when they interact with this persona, they will give them the right context about what's in the pipeline, what is forecasted, what sort of deals opportunities are gonna close end of this month, end of quarter. And it's gonna give them the same preparation notes, right? Like there's, maybe there's a deal that is at last stage, almost about to be closed. Maybe these are the things that should be spoken about to sort of close the deal, right?

So, it's a very simple use case, but the impact is so much uh uh from a revenue, from a productivity standpoint. So I am bullish on this use case. We are trying it out. And it's not just AI, it's AI with orchestration on the backend because for us to get this data, it needs to talk to our CRMs. It needs to talk to our sales system. It also needs to sort of figure out what our goals are internally, right? To sort of see, be like how close, how far away we are. So I think that's like my favorite use case. I would say if we do this in six months, my list would be bigger, right? Like I am definitely bullish for AI across the company.

‍

Evan: One of the areas I've personally underestimated kind of AI capabilities is in I would say in product development, right? Not just code generation, but the entire process and kind of generating, you know, a thing that does something, right? Whether it's an external product or an internal tool. So I have to imagine like the build versus buy equation is changing, right? As the cost of software engineering goes down. 

Um. One thing that I'm struggling with is, you know, where should you be on that spectrum, right? Um. Again, six months ago, I told our team, you know, any AI tool that does anything that's kind of worth it, let's go try to get deployed. And then I was like, wait, no, we should be all this ourselves, right? Because we had to worry about controlling all of our data and there's just too much risk there. And now I'm kind of like 80 % build, 20 % buy, like it's probably gonna change, but also like some of these use cases are so impactful, right? We can't build them all at once. So like, what, what, how do think that balance changes, right? Like how much, like I think it will shift, I think we'd probably both agree it shifts a little bit more to the build side, because just the ROI is higher, but how far like the change over time, you what's your kind of guidance to, you know, some of your executive peers, right, inside the company and maybe your CIO peers outside the company about how they should be thinking about that and how it changes over time.

‍Shyam: I mean, that's, that’s another top of mind question for everyone. The buy versus build idea, right? Uh, 2025, I feel people are at the middle when I talk to my peers because the main thing is you don't want duplication. 

I'll give a very simple example. When you think about collaboration apps, right? I'm not gonna take names here, but our favorite ticketing systems are collaboration systems, for example, right? The question that I hear from my peers is, do I need to turn on AI on two similar products? Or can I just create like a custom GPT that can talk to both of these systems? And now the extension is your enterprise search products, which allow you to build custom apps that can connect to these systems. Is that the right strategy, right? 

So my sort of where I am at this is every renewal that we are going through, we are double, triple checking that, do I need to turn on AI? Do I need to purchase AI offering from this specific vendor? Or is this fall within that custom GPT or custom app category, right? I think that gives like lot of clarity. 

And the other question to ask is, who's gonna be building these custom apps? So if a tool is being used by more of knowledge workers, hands-on folks who can build automation, I think the strategy of building yourself makes sense. But if you have like a ironclad sort of approval tool, for example, right. Along those bucket, I think for those categories, buy would make more sense, right, to some extent. it's figuring that balance out. 

And there's this big confusion now for all, all everyone that should I be paying a premium to open AI's, Anthropics of the world, or should I just like not worry about this, not worry about what's new and just give like a flat fee to my existing SaaS vendors. The other side of the story is this is just me, right? I'm being very bullish here and I might sound a bit controversial. I think the number of SaaS apps in the long run might change in my opinion. And the number of AI agent or AI agent products may go up, right? 

And the other sort of worry that I have or when I talk to my peers is, SaaS sprawl is a thing has been a thing. There might be a new phrase AI sprawl right like that or AI agents sprawl right so you have to sort of think about both side of the spectrum before coming to a decision so.

‍Evan: If anyone said they weren't struggling with this, I would question how much they're really in the details here because every platform is now and it has an agentic tech platform, right? You got it in Microsoft and Salesforce and OpenAI and like, know, like pretty much every layer, like the enterprise knowledge management systems, right? So it's really hard to know kind of like what kind of architectural bets you want to make and, you know, where those are going to pay off and like, what should be localized? What should be centralized? What should be kind of like really like we want to own ourselves first. Like we build on their platform, we kind of get half the work for us. It's really hard to know what the right answer is, but maybe we'll do a part two of this podcast in two years. We'll come back to that and see what, see what we've learned then. 

‍Shyam: Indeed.

Evan: Okay. So imagine we do do part two of our podcast five years from now, right? What are some of the ways you think AI is going to transform how businesses operate that most people would disagree with right now, right? That most people consider, they'd say, hey man, that sounds like science fiction. There's no way we're gonna get there, right? What are the areas where you're bullish that most people are, or where you think most people underestimate the opportunity?

‍Shyam: In five years, CIOs would be doing renewals for AI agents and not SaaS applications, in my opinion. Like that's going to be the future. And the other thing would be five years down the line, you won't be building automations. You won't be doing point to point integrations. You will just give everything to AI. It will figure out what needs to be done and it will give you the outcome and the experience. So, I think that's definitely going to happen. 

Maybe it's five years, 10 years, but the other side is again, wearing my security hat. The trust aspect needs to improve on AI, right? So trust is going to be super important and like how all these players like the open AI is anthropic. I think productivity delivery is one aspect, but security, trust, privacy, like they need to make big investments because the moment trust goes down, things will go in a different direction. So it needs to be a right balance, but if what we are seeing, if the same speed, same direction, same security controls, being mindful, I think that's where we are headed.

‍Evan: If you look at like even the last 12 months, you know, the ability for, you know, a software engineer using modern tools, their ability to kind of increase the altitude at which they're working at, right? If you're using cloud code, I mean, you're writing like documents, right? You're not really looking at code. And so that's happening today in software engineering in 2025. You have to imagine what happens in kind of other areas of applied AI and other departments, right, over the next five years. 

So if we kind of extrapolate that, you know, the cost of software engineering is going down. That's going to affect the build versus buy equation internally. But it also means that like, whoever your top 20 software vendors are today, they're all going to have every product. Right? You know, at some point in the future, everyone can snap their fingers and have every product, right? So every vendor is every product. What happens in that world? Right? 

Presumably there's some consolidation. There's some company, you if everyone is the same product, you're still going to pick some companies for some things versus others. So what becomes those kind of attributes that, that kind of matter outside of product capability, right? If everyone had the same code, why would you buy one thing for one person? Right? 

You mentioned kind of trust, right? You're going to buy from the platforms you trust. Not kind of like you trust the sales rep, but you trust their availability and security and privacy and governance. What are some of the dimensions that become like the big differentiators between vendors and the future where every platform does everything.

‍Shyam: Everyone's going towards a platform offering. Every vendor can do anything and everything with the AI. But again, goes back to the point around business expertise, right? Like that business expertise is going to be core. For example, if you think about like a service ticketing platform, that platform has metadata across different industries, different vertical, and it's not just giving you a very high level wrapper based information, but whatever information or output it's giving you, it's backed by factual data. I think that trust business experience is still gonna hold value in my opinion, right? Because that's what these companies are, right? Like when you think about your CRM, service management, or any like specialized, your financials company, the real sort of meters, they have data, they have metadata, and they are able to sort of influence the journey for a customer based on that. So to answer your point that business expertise is still going to continue to play a key role.

‍Evan: Okay, the last couple minutes, we're gonna do a quick lightning round. So I got four questions looking for like the one tweet response. And these are questions that are very hard to answer in one tweet. So forgive me in advance.  So the first lightning round question is how should companies measure the success of a CIO?

‍Shyam: XLA, how many experiences are they delivering?

‍Evan: What's one piece of advice you wish someone told you when you first became a CIO?

‍Shyam: You have to unlearn to learn new things.

‍Evan: What's a book that you've read, some point in past that's had a big impact on you?

‍Shyam: I have read a lot of books, but the book from my previous company, CEO, The New Automation Mindset, that was pretty, I learned a lot about lot of things. So that would be my favorite book of 2024.

‍Evan: What's an upcoming new technology that you're personally most excited about?

‍Shyam: It's AI, Like AI, but strategic AI, like that's how I would put it.

‍Evan:And maybe what's strategic AI?

‍Shyam: The sales, the agent, AI agent that I was talking about. it's AI, plus context plus automations on the backend. So it's like a mix of three things. So I am definitely bullish on that.

‍Evan: What do you think is gonna be true about AI's impact on the world that most, let's say five years, 10 years down the road, but that most people consider science fiction today?

‍Shyam: I think whatever was not reachable, doable will be actually doable in few minutes.

‍Evan: I think you might be right. Okay, well, it's great seeing you. I feel like we're gonna have some more awesome conversations in future, but thanks so much for making time.

‍Shyam: No, absolutely I enjoyed this. Thank you for the opportunity.

‍Evan: That was Sham Bhojwani, Chief Information Officer at Nextdoor.

‍Saam: Thanks for listening to Enterprise AI Innovators. I’m Saam Motamedi, a general partner at Greylock Partners.

‍Evan: And I’m Evan Reiser, the CEO and founder of Abnormal AI. Please be sure to subscribe, so you never miss an episode. You can find more great insights on enterprise AI transformation at enterprisesoftware.blog

‍Saam: This show is produced by Josh Meer. See you next time!