Shashank Saxena of Sierra Ventures joins Nick to discuss How AI Transforms Services into Software and Revolutionizes Enterprise SaaS. In this episode we cover:
- Vendor Management System Vinley
- The Future of Work and AI
- The Role of AI in Enterprise Adoption
- Regulation and AI Adoption
- Characteristics of Successful Founders
Guest Links:
The host of The Full Ratchet is Nick Moran of New Stack Ventures, a venture capital firm committed to investing in founders outside of the Bay Area.
Want to keep up to date with The Full Ratchet? Follow us on social.
You can learn more about New Stack Ventures by visiting our LinkedIn and Twitter.
Are you a founder looking for your next investor? Visit our free tool VC-Rank and we’ll send a list of potential investors right to your inbox!
0:18
Shashank Saxena joins us today from the Bay Area. He’s a Managing Partner at Sierra Ventures, where he leads enterprise software investments at the early stage. Prior to Sierra, Shashank was the Co-Founder and CEO of VNDLY, a Vendor Management System (VMS) founded in 2017 and later acquired by Workday for $510M. Shashank, welcome to the show!
0:40
Thank you, Nick, thanks for having me. I’m super excited to be here. Awesome.
0:43
Tell us a bit about your background. I know you’ve worked in corporate. You’ve been a startup founder. Now you’re a VC, you know, kind of give us the quick two minute path to Sierra. Yeah,
0:52
sure. Thank you. So prior to this, I’ve had three careers. I started off out of college at Citibank, and I grew up the ranks in online banking, and then the iPhone was launched, and mobile banking came about. So I started off in that team, and actually wrote the early wave of mobility and mobile apps. From there, I went on to Kroger, the grocery retail store chain. So Kroger was starting to build out digital and E commerce. So I got to lead that digital transformation that Kroger built out the e commerce business from zero all the way to a few billion dollars in revenue, and then I worked for the CEO and CFO of Kroger in the strategy division. And from there, I decided to quit and start my own startup. So as you said in the intro, I started my company in mid 2017 raised a seed round. Four years later, 2021 the company was acquired by workday. It was a good exit. So I was at workday the last couple of years after the exit, and I joined Sierra ventures last year as a management partner. So my career pathway to venture was big company, corporate, to start up to Vc. What are
1:52
some of those lessons and principles that you take from your corporate context, which we don’t often have on the show that you think informs your role as an investor at Sierra, I have
2:02
seen scaling in reverse order. Is like, is what I like to think of it, right? So most people go through a startup, and then they see scaling, and then they end up at a big company, and they see scale. Then, for me, it happened in reverse order, where I started at Citibank, which is a massive company, especially in the pre 2008 era, and then from there, Kroger, which, back then, it was number 17 on the fortune 500 list when I was there, and it was 110 100 and $20 billion a year in revenue. So again, massive scale, massive operations, complexity, the value of processes, how things can be done at scale, how money can be made. It’s all lost very quickly at scale. So I got to see that. I got to learn that most companies, most large companies will have corporate bureaucracy, but also, when you turn the ship around and do something, it’s done with the full power of the corporation behind it, right? So I got to see and learn that. And then I went to startup land, where there was absolutely no resources. So I saw the value of both the value of moving fast as a startup, but also the value of doing things at scale from my large company corporate background, perfect
3:01
in your startup that you exited to work day. Do you pronounce that? Vinley or vendley
3:06
friendly? Yeah. So we took two words that generally never go together, vendor friendly, and we put it together to create vendor friendly, which is friendly. And by the way, the domain name was easily available online because the word had no vowels. So that’s why we try to save $10,000 and not have to buy a domain as a scrappy startup, and that’s why vendor league
3:24
so vendor management? Was this vendor management for the enterprise? Yeah. So
3:29
the way we think about it, because vendor management can have multiple different terms, right? So people think of Ariba and Coupa and VMs and P to P and all these different systems. The simplest way to think of us was workday. HCM is the system of record to manage all your employees. Venley was your system to manage all your non employees, consultants, contractors, offshore service providers, all of that stuff. So generally, on an average, 35 to 40% of the Fortune 500 workforce is non employees, which is consultants and contractors. Venley was, quote, unquote, the Work page, cm, to manage all your non employees. So it was that flag, and is that
4:03
how the business started? Did you go through any pivots like walk us through that early path to product market fit and when you found it? Yeah.
4:11
So actually, in four years, there was no time for a pivot. So what we did was we spent a lot of time upfront in creating a thesis around why this space will matter, and is there truly an opportunity? And the founding of the company was four simple bullet points, which was four hypotheses that we put on the PowerPoint slide and we presented it to VCs to then go raise a seed round. And the four hypothesis was first. One was the world of work is changing. When we looked at from the point of starting the company 10 years prior to that, less than 10% of the Fortune 500 workforce was non employees. When we started the company, that number was right under 20% by the time we exited the company, the number was like 33% and depending on which report you read, it’s 40 to 50% somewhere in that range. So the point was, the tail winds are in our favor, where the world of work is changing. More and more of the workforce. Is a rising number of the workforce is going to be non employees. The second hypothesis was the wave of SaaS 2.0 where, if you look at in the enterprise with digital transformation, application modernization, we were seeing this as a trend where workday went in and replaced PeopleSoft and Coupa was taking on Ariba and ServiceNow was taking on remedy. So all of those replacements were happening, and we thought, can we go in and knock out a couple of our legacy predecessors that were born 20 years earlier and not in the cloud? So that was the second hypothesis. Third one was, if workday has a system of record to manage what’s soon going to be 60% of the workforce, which is employees, and has a 60, 70 billion market cap, can we build a company that manages the other 40% and have a decent sized market cap, right? The size of price big enough, and is the opportunity large enough? And lastly, the hypothesis was we would have to build a complete platform, because that was also the era, if you think about seven, eight years ago, there was a lot of gig platforms coming up for work and getting work done. We did not want to build another sourcing platform. We wanted to build a full suite ERP system to manage your entire non employee workforce. Perfect.
6:03
And I would love to talk a bit about the future of work. I can’t I can’t imagine somebody who has more experience than yourself. And let’s start maybe with solopreneurs. They come in many shapes and sizes. You’ve got fractional you’ve got consultants, you could have independent attorneys, right? You could even have people that are gig workers, creatives, and people kind of hustling, starting their own businesses. Talk to us about some of those spaces that you’re spending time in, in where you’re focused on it. Sierra, yeah,
6:35
you know, if you interestingly think about the world of work, and the statement I just made about the changing dynamics of the workforce when I started the company, AI wasn’t a wave right back in 2017 like, of course, people knew about it and they spoke about it, but they weren’t doing as much with it compared to today. So if you look at the strategy before people had two options to get work done. You either hire an employee or go hire a contractor, which is you outsource the work to offshore TCS with pro Infosys on site, Accenture, captcha, minai, Deloitte, whoever right the point was, you either hire an employer, you bring on a consultant or contractor to do it. Now there’s a third option, you spin up an AI agent to do the task. And that’s where it gets really interesting, because now, how do you think about tasks are best done by employees. What tasks you should care about the outcome, but not the process and don’t own it, which is outsource it to a contractor, and for what tasks should you spin up AI agents and automate so you don’t have to, don’t even have to look or think about it, and they should just get done. And that’s really where we think the interesting dynamics will appear in terms of future work, as to which tasks become human tasks, versus which tasks become agentic tasks,
7:47
and where do you net out on that, like, how do you carve up the universe of behaviors and tasks and those that you think humans will continue to be strongest in versus agents or automation? This
7:59
is evolving pretty quickly, but the way I think about it as of today, and that’s why I’m adding the caveat that as of today, because, again, the space is changing pretty quick. Tasks that are well documented, there’s a process guide, there is training that can be provided. So think about handling Center call volume right when you’re onboarding a new call center person, you give them a script, there’s a training guide, and there are certain use cases which account for a vast majority of the call volume. For example, those things are very well suited, and which is why customer support AI agents and issue resolution is one of the low hanging fruit for AI agents, right? So those can get done, whereas certain tasks that are extremely ambiguous, the next set of does not mimic the past set. So historic data is kind of irrelevant in a lot of things, and for those tasks, you’re still going to need humans, because what are you going to train the AI models on? That’s where, again, tasks, where repeating patterns show up, are very ripe for AI agents to come in and automate. I was
8:55
reading an article the other day that was talking about three phases of AI. It was talking about automation, then workflows, and finally, agents. That’s right. So automation is executing predefined rules based tasks. Workflows can call an LLM GPT via API, and then the agents can, kind of adapt autonomously. How far off do you think we are from high performant agents, you know, completing and adapting to complex professional enterprise tasks. I think
9:25
we’re closer than most people think or give credit for so and that’s a little bit of the scary reality where the people who know can see it and will generally hold the same sentiment that I do, that it’s happening now, like in the next 18 to 24 month time frame like it will be pretty ubiquitous. And there are people who are not in, I would say, the early tech Silicon Valley kind of bubbles, and they kind of believe it’s a little more further out than what it is. What
9:52
does soon mean to you? Does that mean one to three years? Does that mean five years? You know what sort
9:57
of time frame I lay out? Yeah, really? Because, I mean, think about it, this conversation has been having for the past 18 to 24 months, also, right? So in the four year horizon, if you look at any of the waves, let’s take mobile the iPhone was launched. I’m guessing what 2007 2007 Yeah, and 2008 was the iPhone 3g with the App Store launch. And by 2010 it became pretty ubiquitous. Now, what really helped is the fact that AT and T subsidized the iPhone and made it a $200 device, and people didn’t have to pay $1,000 up front for it, right? So of course, that helped the adoption. But the point is, within three to four years, it was mainstream. It was no longer a niche technology, a cloud wave, if you think about it, right? That took almost six years in the enterprise sector. But so the fact that we’ve been speaking about it for two years, and I’m giving it two years more, four years, is a pretty reasonable time frame for this to go mainstream. So,
10:54
so let’s pull that thread a little further and extend the analogy. So if, if you know, the App Store was the big winner in developers building apps in the Apple ecosystem versus Android, or Microsoft was trying to do the Windows Phone, et cetera. What do you think in the LLM world, whether it be cloud or llama or open AI or even deep seek, what becomes the de facto platform that the majority of developers build the most on,
11:23
I don’t think it’ll matter. And here’s why I say that, if you think of and I’ll share a little bit of the VC perspective, but again, this is just my opinion, right? So if you look at the cloud wave, and you break it down into three separate layers, right? The infra layer, like that’s not where Vc returns were produced inside big tech ended up winning that between Microsoft, Azure, AWS, GCP, like the cloud providers and hyperscalers. The data layer above it, there were a few winners, Databricks and snowflake and a few others. But a lot of the value was created in the Applications layer, in the cloud wave, Salesforce, workday, ServiceNow, Coupa, there’s a whole list, a long list, of winners in the application. Similarly here, when you look at the AI wave, which is the next big transformative wave, the infrastructure layer is very, very hard to invest in, because we think again between the big tech players, the same names, AWS and Azure and GCP plus and a few others. Those are all public companies, right? Like VC returns are hard to drive. The layer above is the models layer. So today, when you say AI company, most people think models and that, I think between anthropic, open, AI, between all the stuff going on there, I think that that is very hard to be investing in, but the layer above is what most people generally don’t talk about, which is the apps layer. And you’re starting to see a few winners in that, like Harvey and legal tech and few others. But the point being that apps layer is, I would say, under invested in is going to drive great VC returns, and that’s where I would say the next wave of winners will likely come from.
12:56
It feels like people are finally talking about it. I mean, I’ve been preaching to my LPS about it for two to three years, and phone has blown up in the past month or so with, oh my gosh, everyone’s talking about the application
13:06
layer now. Yeah, exactly in fact, in December I’d done in a pitch book article, I made a comment around how the infra layer and the models layer is over invested, and the apps layer is under invested. And I had a couple of people comment on it and comment that they don’t agree with the point of view. Interestingly enough, the deep seek release came out in mid January, and everyone’s leaving. It’s like, okay, infra and models might get commoditized at some point, and this is no different than the internet days, right? Cisco was the most valuable company, but then the applications company that were built because Internet became ubiquitous, that’s really what happened in the.com transition, and then the application companies after that. So similarly here, where infra, GPUs, models all that’s great. They’re the fundamental underpinnings, and they’re absolutely needed and necessary. But what happens and what’s built on top of that is going to be the next wave of value unlock. Perfect.
14:03
So I think I know where you’re going to net out on this question. But is, is the deep seek news a net benefit or a drawback for
14:10
the application? I don’t know. We’ll see like this one’s the fact that SlMs are more efficient than llms. It’s like, it’s not a surprise. Like that should not surprise anyone. The key is now, what do you build once you have access to models that are reasonably priced and efficient? It’s a little bit like the iPhone example you’re giving, right? When once you have Google Maps and a phone advice on every person out there, you can then go build Uber, because without that, you can’t go build Uber. And similarly, here, what can you build? So this whole terminology of AI company, and we did the same thing in the cloud. Wave of cloud companies, where workday is a cloud company, for example, right? Or Salesforce was built on the cloud today, when I start a company, or when I started, when leaving in 2017 What’s the definition of cloud companies? Obvious that if you’re starting the next startup, it is going to be on the cloud. What else is your option? Are you going to go buy servers and set up your own data center like that’s not an option. So just the de facto meant, of course, you’re a cloud company, of course you’re hosted on AWS or whichever hyperscaler you prefer. But the point is, it’s a de facto, right? So today, the terminology is, when you say AI company, there’s an inherent assumption that you’re talking about like models. But that’s not necessarily true, and that’s changing pretty rapidly, where AI company can also mean apps. So AI company being modeled, AI company can also mean apps to then AI company is obvious, because every company does AI in some way shape or form, just like most companies are hosted on the cloud in some way shape or form. So that terminology is, I think, going to phase out and not age gracefully over the next couple of years.
15:41
Well, and you know, as deep seek or other models get more performance, and supply side of fuel, or whatever you want to call it, at the model layer drops precipitously. I mean, this is kind of the definition of disruption, right? When supply side costs drop, then demand, you know, increases is that, is that fair? That’s
16:03
right, because more people start doing more interesting things with AI. I mean, if you look at AI today across the I would say legacy, fortune, 500 landscape, keep aside your cutting edge tech companies that do tech for a living, right? But if you look at Midwestern let’s take banks or insurance companies an example, they’re doing stuff with AI, but I would say they’re more experimenting AI than have mass market adopted AI. So it’s still lab projects, it’s still proof of concepts, it’s still pilots, it’s still a handful of engineers playing around with it. Some use cases, some interesting use cases that they’ve automated. They’ve not done big, massive enterprise rollouts where AI is the way they’re running the business, and that I feel is going to change, and that’s the corner that’ll turn in the next couple of years. As to your point, costs become cheaper, things become more reasonable. Adoption goes up, and developers build more interesting tools that address more meaningful use cases. Back to
16:57
where we started at the top with venley. Do you think a lot of these large enterprises that are not tech enterprises are going to be able to attract the Forward Thinking AI talent internally to do this? Or do you think it will be vendor management thing where they’re bringing in the best consultants, pairing that with the best you know, external technology.
17:21
The point is, they will not have to. So think of the cloud wave like, how many companies went ahead and moved their data center and ran their own cloud instance, versus just using it out of the box from AWS, for example. And that’s a little bit of the point with AI is, once it becomes cheap and ubiquitous, they’re not going to have to go hire armies and armies of AI engineers, because today’s engineers, the tools that they’ll be using, will have aI baked into them. So you won’t need a specific AI engineer unless you’re building deep tech or models, or any of those, I would say, infrastructure layer components. How do you advise portfolio
18:01
company founders that have built on more of a SaaS Foundation? And you know, maybe they’ve progressed through some series of funding and some growth and and now they’re kind of at this crossroads where they might have to reorient the ship or the aircraft carrier, because the the foundation of the way they’re approaching things is, it was not AI first, and now we’re living in a different time. Yeah, the three
18:25
paradigms that I would think about if I was still a founder operator is intelligence, automation and user interface and user interface, they’ve had to live through couple of other generations, right? Websites, mobile apps, every unthought websites will die because everyone will do everything on mobile. Didn’t really happen. It became an omnichannel world, similarly for AI agents, right? I don’t think the applications and mobile apps are going away just because there’s an AI agent. They may make it more efficient. Might be one more channel. But the truth is, we live in an omni channel world, right? When and Siri came about. Voice was supposed to be the killer interface, and voice was going to take over everything, right? It’s like that didn’t play out exactly that way, either. So when you think about the user interface, it’s like, what a right task for which you need an AI agent is a better interface to handle things as compared to the user going in and doing things. And this is where, like, push versus pull tasks in terms of things that are done, and you’ve just been notified that they’re done, versus things that you need to go and take the trigger action or trigger an event to go do things right? And that’s where, when I talk about these three interfaces, they all go hand in hand, which is automation and user interface
19:43
Perfect. Let’s go back to to workforce in future of work. What do you think happens to kind of the working economy 10 years from now? Where are the majority of jobs going to shift? Where do we see contraction and where do we see expansion in the future of work? It’s
19:58
interesting. So far for the 20 or 30 years. So this is pre AI, even pre internet. Honestly, the overall sentiment was the light industrial, blue collar kind of jobs to get automated, right, either with robots doing it, machinery, equipment, depending on which wave of the Industrial Revolution onwards, you think about like the whole thesis was around automating the blue collar workforce. Now the risk, for the first time, is for the professional workforce. And this is where, when we think about SaaS, SaaS is not just software as a service. The new paradigm around it is services as software, which is, how can you run service firms, accounting firms, legal firms, law firms that are incredibly efficient, that use AI agents as your as a digital worker to get things done. So that’s where the ratios will start to go off. So far, at least in the last couple of waves the there was a lot of counter intuitive reward system. So for example, if lawyers make money on billable hours, making them more means less billable hours, which means less revenue. So why, again, is efficiency a good thing for them, whereas now I don’t think it’s going to be a choice. So this is where, when you think about services firms and professional services firms that make money on billable hours, those are the ones that are going to get severely impacted by AI because, and this, by the way, also falls into my theme of employee, contractor and AI agent, right? So if you think of agencies, creative agencies, to create a design mock up or to create a campaign, you could have hired employees. You could have outsourced it to an agency for contractors. Now you can spin up an AI agent to lock ups and designs. And by the way, the outsourcing piece, they have couple a couple of iterations of it. One was a traditional agency. Then came Fiverr and these gig market bases where you could go hire someone that was the contingent worker bucket. And now you can create stuff pretty well with AI agents, including things like podcasts or mock ups or videos or and we see a lot of the tooling that is provided. The tech is getting there pretty quick.
22:10
I think the thing that concerns me a bit is humans. I guess the human race is pretty resilient. It evolves pretty quickly. But a lot of humans don’t like change, and they like to learn a job, and they like to do it over and over again. And in a very sort of general sense, abstract sense, AI is really good at doing repetitive tasks much better than humans is are. And so my fear is that it’s going to force humans to adapt and change and learn and be more creative on a regular basis, and there’s going to be a group of humans that do that and kind of continue to evolve, and there’s going to be a lot of people that that don’t and don’t want to and are going to be resistant humans,
22:49
to some extent, lazy and will always try to find the path of least resistance, which is the most efficient way of doing things, which also takes us all in the different curve of technology adoption, but at the same point, there’ll be early adopters that’ll move sooner, and then there’ll be Yeah, so again, we’ll go through a wave of that. So it’s not going to happen overnight, but it will happen, and we’re going to be on different points of that journey, depending on where you fall on the curve and how tech savvy you are. With
23:18
regard to AI Shashank, do you think we need more regulation or less? I guess we don’t have much yet, but do we need more regulation?
23:26
I would worry about regulation coming too early, so I don’t worry about more or less. I think it’s about timing, like let the tech evolve, see what happens, and then start to regulate based on having a use case in mind. The problem is today we don’t know exactly how it’s going to be used. So if you don’t have a problem statement and you’re just trying to regulate for the sake of regulation, that’s a pretty dangerous place to be. So you have to understand what is the problem statement you’re trying to solve for for which regulation is the solution and regulation is not the first line of action. Regulation should be the last line of defense. In terms of, okay, without regulation, this will never get done. Hence, we put create regulations for this. So I do believe in having guard rails, but also today, it’s so early where we don’t know what’s the problem we’re trying to solve for and putting regulation in place will only slow down growth, and if we regulate and other countries don’t, we will be behind in the AI race, which I don’t think is the right way to go. So that’s why I worry about regulation coming in too early. So this is not about more or less. It’s about regulating it too early will be the problem
24:37
Shashank. Do you think successful AI founders are going to be fundamentally different than successful founders from previous platform shifts or eras.
24:48
No, I do think the same key and core tenants of entrepreneurship apply to every generation of founder, irrespective of if you built a company doing a.com boom. The cloud era, the AI era, there are certain traits around founders that are just core human nature and human elements that are needed to be a good founder, irrespective of the era in which you’re building a company in.
25:16
You know, how do you screen for those? What are you looking for? You’ve been a founder yourself. I’m sure you’ve done some reflection, like you hold up the mirror and you see a lot that you like, maybe some things that you don’t look for. But you know what, if you had to boil down, like, those key characteristics or traits that are immutable that you’re looking for? Sierra, what are those?
25:35
Yeah, and this is a tricky one, right? Because we all have our own biases in terms of what we think makes a great founder. And then venture land, we try to pattern match things which may or may not be a good thing, because founders come in all shapes and sizes, right? Like they are, they come in all form factors. You have some who are brilliant jerks and some who are absolutely great, good to Earth, nice, humble people, and you have a whole spectrum all the way in between, and trying to pattern match a stereotype is always tricky in that regard. So but the core tenants are around grit, determination, resolution, the ability to handle the extreme highs and the extreme lows of the journey without it getting to you, or even if it gets to you, the ability to vent it out with a core group of key, close people that you’re you’re good friends with, either board members or your personal life friends, right? So all of those things, I would say, are building blocks of foundation components that are needed in the founder journey. Because this journey is, it’s stressful, it’s emotional, it’s fun, exhilarating. So you can use a variety of contradictory adjectives to describe it, and the contradiction is what makes it really interesting. Actually love it.
26:56
And you know, finally, I would love to hear kind of what sectors or industries are you leaning into most right? Lots of changes in technology, lots of opportunity. Are there certain categories or sectors that you’re spending time in? Yeah.
27:11
So at Sierra, again, we are enterprise software, which means everything right from cyber security, infrastructure, Dev Tools, applications, vertical SaaS, like we do all of it, personally, I focus on more of the apps layer, so between applications and vertical SaaS, like those things are things I understand well, anything that would be selling into the persona of a fortune 500 CIO is generally the tool set that I understand well, having from that World of large corporate
27:40
Shashank, if we can feature anyone on the show, who do you think we should interview, and what topic would you like to hear them about? Oh,
27:47
you know who you should interview is my mentor, and he’s played the role of a father figure in my life. Steve Singh, who is founder of Concur which was acquired by SAP for whatever, 8.3 billion. He’s a managing partner at Madrona, but the early days of cloud and how he took an on prem company to the cloud, and how did Wall Street respond, because it already a public company, then he’s got some super great stories and insights on that. Shashank,
28:13
what book, article or video would you recommend to listeners?
28:17
Recently, I was very impressed with Brian Chesky is interview with Keith rapiers in terms of, I mean, it was a whole Airbnb journey. So they touched upon a bunch of different topics, including VCs members, how to pick a board member, founder mode versus manager mode. So there was a lot that was covered on it. I really, really enjoyed. It was, I think, an hour long segment interview, but I really enjoyed that one,
28:43
perfect Shashank. Do you have any habits, tactics or techniques that are a force multiplier? Slow
28:49
mornings? That’s trick is most people wake up rushed and they feel they’re caught behind and the list is too long and the day is getting away from them. I take my mornings super slow, and that just helps to start the day with the right mindset. I’m in the right mental state of mind, and I feel good about the whole rest of the day because I don’t feel rushed for that first 30 minutes or an hour in the morning, perfect.
29:15
And then finally, here’s Shashank, what is the best way for listeners to connect with you and follow along with Sierra,
29:20
add me on LinkedIn and ping me. I respond to most pings I receive on LinkedIn. He
29:25
is Shashank. Texena, the firm is Sierra Shashank. Thanks so much for the insights today. This was a lot of fun. Thanks. Think
29:32
it was great chatting with you. Thank you. Thank you, sir.
29:40
All right, that’ll wrap up today’s interview. If you enjoyed the episode or a previous one, let the guest know about it. Share your thoughts on social or shoot them an email. Let them know what particularly resonated with you. I can’t tell you how much I appreciate that some of the smartest folks in venture are willing to take the time. And share their insights with us. If you feel the same, a compliment goes a long way. Okay, that’s a wrap for today. Until next time, remember to over prepare, choose carefully and invest confidently. Thanks so much for listening.
