Ep. 18 Ross Mason on the Power of "Headless" Applications

Decoding APIs: Ross Mason on the Power of “Headless” Applications


29 min

Ep. 18 Ross Mason on the Power of "Headless" Applications

29 min

You can't see them, but APIs are always hard at work, sending and retrieving the information that makes the digital world possible. When our data stops flowing, that's when the trouble begins. In this episode, Ross Mason, pioneering founder of MuleSoft, explores the importance of APIs and why connecting data from many sources is one of the toughest problems companies face.

Read transcript

“Data is better than fuel. Fuel just makes something go faster. Data creates this new asset class of things that you can go and create brand new things that are even bigger than the thing that where the data came from in the first place.”

Quick takes on...

Defining APIs

“You hear people say an API is a product. The product realization of an API isn't building the API. It's putting all the infrastructure around it to help people get successful using it.”

API-Led Connectivity

“We’ve barely scratched the surface of what connectivity really means. APIs need contexts that know how to connect better and more easily, and group things together in a way that we can understand better as humans.”

The New Role of the CIO

“The role of the CIO is really to figure out the strategies within the organization to make [data] broadly available and actually get those assets to the front of the business where they can be used most effectively.

Meet your guest, Ross Mason


Spotlight on Ross Mason talking about "headless" applications

Ross Mason is the founder of Dig Ventures, a VC firm that makes early-stage investments in B2B enterprise SaaS companies. Prior to Dig, Ross was the founder of MuleSoft, a company he launched in 2006 based on the idea that connecting applications should be easy. While there, he was responsible for MuleSoft's product strategy, open source leadership, engineering alignment, and direct engagement with customers. He also led MuleSoft through a successful IPO in 2017, and then a later acquisition by Salesforce in 2018.

Listen to the next episode

Ep. 19 Jim McKelvey headshot discussing the Innovation Stack

Decoding Innovation: Jim McKelvey on "The Innovation Stack"


37 min

If you think great innovators have superpowers, Square co-founder, Jim McKelvey, will make you think again. After Square survived a direct attack from Amazon in 2014, Jim was determined to find out why. In this episode, learn what it takes to build a world-changing company, and why solving difficult problems is the true key to creating resilient businesses.

Episode transcript

Ross Mason:  [0:06] Data is better than…

Ross Mason:  [0:06] Data is better than fuel. Fuel just make something go faster. Data creates this new asset class of things that you can go and create brand new things that are even bigger than the thing that where data came from in the first place.

Dan Saks:  [0:23] That's Ross Mason. He's the founder of MuleSoft, a pioneer in the API platform sector. Ross is a giant in the data world. He was among the first to recognize the potential of big data in the enterprise and wrote the first line of code for MuleSoft in 2003.

[0:43] In 2017, he led the company's successful IPO, and the next year, he led MuleSoft through its acquisition by Salesforce for $6.5 billion. Ross left the company in 2020 and went on to found Dig Ventures, a VC firm that makes early stage investments in B2B enterprise SaaS companies.

[1:08] Ross also has tremendous insights on entrepreneurship and how to build companies that last. In this episode, Ross talks about all that and more, explaining why connecting information is so critical. Why CIOs are often prime candidates to become CEOs, and why it's important for founders to find balance.

[1:31] This is Daniel Saks, Co CEO of AppDirect, and it's time to decode the power of APIs.

[1:42] Welcome to "Decoding Digital," a podcast for innovators looking to thrive in the digital economy. I'm your host, Daniel Saks, and I'll sit down with other founders, CEOs, and change makers to decode the trends that are transforming the way we work. Let's decode.

[2:05] Ross, great to be connecting.

Ross:  [2:06] Hi, Dan. How are you?

Dan:  [2:08] Amazing, and you?

Ross:  [2:10] Very good. Thank you.

Dan:  [2:12] Great to be catching up with you. Congrats on the MuleSoft acquisition a few years ago by Salesforce, and then congrats on starting Dig Ventures. Tell us a little bit about what you're doing now with Dig.

Ross:  [2:24] While I was at MuleSoft, when you build a company, you end up stuck in this bubble that you created that everything that you interface with is things that you've created or has come out of your ideas. I started meeting founders for coffee, because I have coffee every morning, for 20, 30 minutes in San Francisco.

[2:46] After a while, I met a lot of founders. There were some I really liked, and I started angel investing. I did that up until 2019, I think, so six years. When I moved to Europe, I thought I really enjoy investing. I love spending time with founders. I love hearing about how people are trying to solve new problems.

[3:08] The problem landscape keeps changing. It's so varied that it's a super interesting time to be in tech generally. I thought I'd start a fund in Europe and work with European founders, bring some of the knowledge that I had from the US over to Europe.

[3:25] Dig stands for two things. One, I've got to like what we invest in. That's why it's called Dig. If I don't dig it, then I don't invest. You want to be talking about the companies you invest in at any social gathering or anything like that. Sounds a little bit weird saying that now because no one's been in social gatherings for a while.

[3:43] The other thing I wanted to redefine what value add was for founders. All investors say they value add, and the reality is I think the mileage varies greatly with the VCs and the seed funds. What we've been doing this year is thinking about how can we add a lot more 10x or even 100x more value than people are used to?

Dan:  [4:05] Let's rewind to the beginning. I know that you were an IT professional originally, which led you to found MuleSoft. Can you tell us some of that initial pain point back in the day?

Ross:  [4:15] I don't know if people would have called me a professional back then. I was more of a cowboy.

Dan:  [4:19] [laughs] I like it.

Ross:  [4:21] I worked in software during Y2K, that was a bust. I thought it was going to be amazing. I worked a lot in banking. Not because that's where I wanted to be, it's just where all the high paying jobs were.

[4:33] I started realizing that, A, they're very complicated. Loads of layers of stuff all over the place. A lot of manual, a lot of different things for different needs. Nothing's really connected. Nothing really talks to each other. I ended up gravitating from apps to integration.

[4:53] My first big integration project was in London for a bank called RaboBank. It was incredible because we connected seven systems, which is nothing. Seven systems. There were about 12 different teams across all these different systems who had different agendas, and there was a core integration team.

[5:13] We created an architecture. It took 18 months. We did barely any testing on it. We never tested the disaster recovery thing that we created for it. We went live, and it kind of worked. You could scale it. I don't know. For me, it felt like it was a miss. Actually, for everyone else around us, they were hailing it as this great new way of doing things. It was called an ESB, Enterprise Service architecture.

[5:38] I just looked at it and I thought if this is what people expect, we can push the bar much higher. That got me thinking about, OK, what would I want to do this differently as a developer? That's where I started.

Dan:  [5:52] Powerful. A few years back, you wrote a book about the IT professional and how you have to break it. Can you tell us a little bit about that?

Ross:  [6:01] The book was called "First, Break I.T." It wasn't like a Gene Kim book. I love Gene Kim. He writes amazing books. He puts a lot of thought into it. We talked about collaborating a bit.

[6:13] In the end, I realized this book wasn't supposed to be a Gene Kim book. It was supposed to be more of a manual. Not a heavy manual. A light manual to get, especially IT people who are managing, directors, or execs, on the same page around some key things, which were that the way we deliver projects in IT is fundamentally broken.

[6:36] Someone like a Unilever will do 250 to 500 IT projects every year. They reinvent the wheel every single time, not because they're stupid, or because the people there are stupid, but just because it's such a large organization. There's no shared assets of any description. About the only thing shared is the vendors that people use.

[6:58] IT has grown up, basically standardizing on vendors, but not standardizing on approaches when it came to IT. It's still very much like this. Our big idea was IT is a cost center. The budgets for IT go down every year. Even though it has all the digital assets, which if you think about it, after people, is the most valuable asset.

[7:22] In a lot of ways, people are temporary. They don't stay in a company forever. Your digital footprint is a treasure trove of everything that you've ever done with everyone who's touched your organization. Yet the people who manage that and hold that, treated it like a cost center. Why is that?

[7:39] It's because they don't get any reuse out of it. There's no way of discovering the information in a reusable way that more than one party can get access to. What we started creating was a methodology for unlocking more of these assets, and a methodology for creating projects around those assets so you got more leverage as you worked a lot.

[8:00] The book was essentially just a tip to you got to stop doing things the way you're doing them. There's a wholesale IT operating model change. It's not massive, but you have to create some new teams, you have to stop doing some bad habits. The book was there to spark that conversation and get people moving more that direction.

Dan:  [8:20] I think it came out at a relevant time where we're in the midst of big digital transformation. A lot of companies and CIOs were determining, do we go all in the cloud? What does this consist of?

[8:30] I think now, particularly this year, everyone recognizes, from the board to the C suite down, that there's a critical need to operate digital first, and where IT is not just supporting the digital or technology function, but supporting the business processes and operations. Can you tell me how the role of the CIO needs to evolve to meet this current demand?

Ross:  [8:52] That was one of my chapters. I'll refresh it. Actually, it's changed quite a lot.

[8:56] If you subscribe to the idea that data is an asset and it's like a treasure trove. It hasn't yet been sorted out and figured out where the value is, but it's there, then the role of the CIO is to figure out the strategies within the organization to make it broadly available, and get those assets to the front of the business where they can be used most effectively.

[9:18] To that regard, you are starting to see a shift where more CIOs are now part of executive boards, real boards of the company, which I think is super important. Otherwise, there's nobody there speaking that language, helping people understand why things take so long and what kind of changes need to happen in order to speed things up.

[9:36] You're starting to see a few people transition from CIO to CEO. I love that idea. That works if the CIO is operationally minded and is thinking around running their capabilities as a set of operations as a service to the business. It's very natural for that CIO to move into the CEO role.

[9:55] Interesting, what has to happen between the IT and the business is IT has to let go this idea they've got to build everything themselves. They do that because they haven't put the infrastructure in place to allow other people to get successful with the assets that IT controls without making a mess, or without creating security breaches, etc.

[10:15] The business also needs to have a more directive strategy on how they build their development and data science teams in order to take advantage of the assets that are available to them in the organization.

[10:27] It's also very easy for the business to say, "Oh, we need these things. We haven't spec'd it out very well, but we've thrown it over the fence of IT, and they've taken four months to go figure it out.

[10:38] That back and forth in the business creates conflict. It makes the IT look bad, but in reality, the business also needs to get a better understanding of their assets and what they need, and how to specify them. Both sides have to come together, but it's definitely the CIO that needs to move first.

Dan:  [10:56] When you see the role of the CIO evolving, how do they collaborate to think about the business requirements and benefit from the assets, which is the data?

Ross:  [11:06] Back when I was active at MuleSoft, it was mostly around getting agreement between IT and the business, and who owns what piece of the puzzle. For example, the IT organization would own SAP absolutely.

[11:24] If you built APIs on SAP, you'd also own those because they're very tied to the application. Those APIs, to get a bit technical, aren't very useful for people in the business because they expose all sorts of SAP junk that nobody else understands unless you understand SAP.

[11:39] We used to have this API led model which is three layers of APIs. The first layer is to decouple the backend system with an API so you can access it through newer protocols to make it easier for new tools and programming languages to connect to it.

[11:56] The layer above that is a set of APIs that have been packaged up within the context of a business unit. Rather than exposing 500 fields in an SAP object, you may only expose 15 to describe part of the resource that it's modeling, plus you might take a few data pieces from other locations. That resource becomes the API that people will build applications and work with.

[12:23] Not revolutionary, but just being systematic and thinking about how to break systems down into APIs so it can be more reusable was a big thing.

[12:32] Now, IT can go and do that, but what happens is IT then says, "Well, I got to run all this stuff. I can't afford to run all this infrastructure." The reality is as soon as those APIs become business context centric, it's better for the business to own those APIs.

[12:48] They can have more control over the life cycle on the specializations on these types of other things they bring in. It's better for a business unit to have more control over their second layer APIs than the first layer.

[13:01] There was always this dance of helping the business understand, "You've got to fund people in here as well to make this work," because they like buying applications. The reality is all in APIs is a headless application, but it was hard to get the business to understand that. I think they're starting to understand it a bit better.

Dan:  [13:18] Let's decode this a little more. For the listeners that don't know what an API is, can you explain what it is and how they can get benefit from it?

Ross:  [13:25] Yes. An API is, essentially, it's a contract between a supplier that has a digital asset or capability and a set of consumers that want to use that asset if it's a piece of data or capability.

[13:40] The API is the programming way of describing how you connect to it, how you query it, and what information you'll get back from it, and when things go wrong, and what happens. It's very basic. It's just a I/O mechanism for the digital economy. The digital economy is a sum of all these interactions through APIs.

[14:02] An API, honestly, can be protocol agnostic. People get very around the axle on it needs to be X or Y. It doesn't matter.

[14:12] What does matter is that it's described well, that it covers the cases so somebody can discover it and use it. That it's put somewhere where it can be discovered and there's someone in the organization or a group in the organization that knows how to help people when they get stuck.

[14:29] You hear people say an API is a product. The productization of API isn't building the API. It's putting all the infrastructure around it to help people get successful using it. One of those is self serve, but quite often in the enterprise, self serve isn't that normal. You also have to have these enablement teams that work with the APIs to help people get the most out of what's available to them.

Dan:  [14:52] Got it. Could you say it like, "When you think of oil to drive your car, you [laughs] need pipelines, you need gas stations, you need infrastructure, and then there's a connection point," and essentially, what data is is oil and the pipelines are the interfaces?

Ross:  [15:08] I knew you're going to go there. I'm going to say, no, data is nothing like oil.

Dan:  [15:12] Is it still valuable?

Ross:  [15:14] Well, oil is just a commodity, and it goes up and down based on demand. What makes data much harder is it's different to different people, and the shapes of data is different in different parts of the organization.

[15:28] It's interesting when you draw a map of what the organization uses, and you've got, say, 30 applications. Everything, from Salesforce and Slack    well, that's Salesforce now    and everything else in between, all of their apps, they're apps. In fact, the way their organization uses them and how they come together on them and what kind of information they want from them is totally different.

[15:51] All those different pipelines to those apps are not treated equally, and they're not of the same value. What you have to do is normalize access so the data pumps. The first layer is to make the data pumpable. The layer above it is then to specialize that data so people can get the most value out of it.

[16:10] Amazon had to go through this. There's a very famous edict from Bezos I used to like talking about which was he told every team that they had to interact with every other team through well defined interfaces. They couldn't use backdoors. They couldn't do it any other way.

[16:27] What they created was a very loosely coupled environment of software. It also created a lot of repetition because they didn't put any rules around. I heard, reportedly, they had about at least 60 different objects to represent an invoice, because invoice meant so many different things for some of the different parts of Amazon.

[16:43] But in the midst of that chaos, what they also have is all the building blocks to go and build everything.

[16:49] Data is better than fuel. Fuel just makes something go faster. Data creates this new asset class of things that you can go and create brand new things that are even bigger than the thing where the data came from in the first place.

[17:06] It's this very abstract but very exciting element in our lives today which is how well can you find the right data sets, put them together in interesting ways, and, everything from the way you store them, to the way you present it, to the way you report on it, changes the outcome. It's almost like some element we've never seen before. It's much better than oil.

Dan:  [17:27] What are examples of businesses who are effective in using APIs, and what are examples of businesses that maybe a struggle?

Ross:  [17:34] I would say a lot of businesses at this stage are doing a reasonable job, partly because if you've gone to the cloud, the way the cloud is connected is through APIs. As soon as anyone connects anything to anything else in the cloud, it's using APIs.

[17:48] Luckily, we went there by default. Thank goodness. We went there by default because if you don't have an infrastructure, there has to be another way to get at the data other than through the user interface. Every cloud provider has APIs.

[18:02] That's made things easy, and you think, "OK, is there opportunity to do more here?" Has it waned or is it we hit the peak of what APIs can do? What interests me, what we haven't figured out how to do yet, is stitch together hundreds of APIs very easily.

[18:19] MuleSoft allows you to do it, but what we found even the biggest implementations can get really complicated because of all the interactions and the way it works. You can do it, but it requires teams of people.

[18:32] The other thing that still needs to be solved in APIs, even as they are today, is monetization. We're horrible at...Especially data APIs. The reason why there's not loads of great data APIs you can go and use is partly to do with that shape problem that one person wants data a certain way, one wants it somewhere else. It's hard to capture that in the way APIs are defined today.

[18:55] Also, they're very, very hard to price. People don't know what the data is worth. They can't model it, so they often don't make it available.

[19:03] Then, finally, it's trust, and I think this is the big one a lot of people are trying to solve right now is can I share data with someone without giving up the inherent value of the data? By virtue of me giving you my data, I never get that back again. There's no subscription model there except for updates.

[19:20] Well, that might not be interesting. If I've given you five years' worth of data, that might be enough for you to go and build your own thing and you'll never need me again. The data monetization piece is still super nascent.

Dan:  [19:31] Earlier in the conversation, you mentioned the power of headless experiences, but can you speak a little bit to what you think the future is for the way that people interact with their applications?

Ross:  [19:42] Yeah. The first thing is the way I look at the industry, at least I have done the last five years but I think it is valid for the next 20, the recipe for success is connect, access, and then enable.

[19:56] First, you have to be able to connect to the things that matter. That's pretty much what we went for MuleSoft, just being able to connect everything. Amazingly, as big as we got and as big as it's getting under Salesforce, we barely scratch the surface of what connectivity means.

[20:12] To me, even with an API, it's super hard to connect, so you want things to sit on top of APIs that have context, that know how to connect better more easily, and group things together in a way that we can understand as humans better. Then, as machines, it's a whole different game. We can talk about that separately.

[20:29] The future of apps as they are right now, I've been investing in certain apps the last three or four years that are removing the need to put data in, and they go and discover the data, then using a lot more ML to scoop information up.

[20:42] Frankly, we've been doing it with Gmail for five or six years now where you give certain things you trust access to Gmail. That sucks it up and discovers things.

[20:51] The first thing that blew my mind that did that was TripIt. I don't know if you remember TripIt, but I traveled a lot, and this thing was sucking all my travel itineraries and make sense of what I was doing on my next trip. That's a great early example. It's a bit like what a Zipcar was to Uber in some ways. It was too early for its time. They didn't develop it any further than they got it.

[21:13] The new interfaces will be much more conversational, whether that be voice, whether that be through text, or whether it will be based on the things we do.

[21:25] Right now, phones have a ton of understanding about what we do, where we go, what we do, and yet the best they can tell me is, when I go get a coffee, that's going to take me five minutes to get home. It's like, "All right, but I've got my calendar. I've got all these other things."

[21:41] It doesn't tell me, "Oh, you have a podcast in 20 minutes and you're 15 minutes away from home. You futz around with your setup for the first 5 minutes, so you should probably rush back."

[21:53] I think the biggest thing missing for me right now, and where I get the most excited, is when people start telling me about how they use the context of action and the context of behavior in order to do something new. It's been difficult up till now because it hasn't been accessible.

[22:10] Because we've connected a lot more, things become more accessible, and then the enablement piece is smart companies that go and figure out how to join the dots between all this noise. Pick out the signal that matters to create new valuable services and products, or ways of doing things, etc. That for me is pretty exciting.

Dan:  [22:29] What's some parting wisdom for IT professionals out there that want to thrive in the next 10 years?

Ross:  [22:37] Oh, all right. Well, given how many pitch decks I see that come through the transom, many of those people are going [laughs] off and starting to build their own companies.

[22:47] I will talk to the founder people first. I spend a lot of my time speaking to founders, and everyone I've spoken to basically say this. Looking after your mental health as a founder is super important, but that applies to anyone.

[23:02] It's just that you put unrealistic goals on your head, and also investors do, etc., when you become a founder. You basically say, "I want to go and change the world," and then you realize, "I'm just me and two friends." That's really stressful.

[23:16] Looking after yourself a bit more than working 100 hours straight every week for five years, it's a good way to go and build a company, but it's not the best way to live. You end up sacrificing other things.

[23:30] For professionals in general, what's interesting in the market right now, and where I like to invest a lot, is in no code platforms. What's interesting with the change we're going through now is you don't need to be able to code to go and build a real product, and that's pretty interesting.

[23:47] We're going to see a different breed of startups, side hustles, everything else, where people don't go for the big billion dollar unicorn, but they build these tools that they can plug into other ecosystems very easily, and some will do well with it and some have side hustles.

[24:04] For an IT professional, the thing to do to stay relevant is to pick one or two niches that you like and stay on top of those versus trying to keep abreast of everything that's going on, because it's just so broad that it's too hard to keep track of everything. Then, reading about funding announcements on TechCrunch is probably not the best use of time either.

Dan:  [24:25] You made a good point around how not everyone has to start to create a billion dollar company. There was a perception in the early days of tech that if you started a tech company, it had to be massive.

[24:35] To your point on low code/no code and other basic businesses that can be upstarts, whether it's digital commerce companies selling things online, there's so much potential for people around the world to start a company.

[24:48] It'd be a great lifestyle business, generate cash flow, and then [laughs] balance your mental health as well, and just living a great life holistically, right?

Ross:  [24:56] Exactly. The cost of starting up is super low. The building blocks are essentially free. You build on AWS or Azure, or Google as a third option, but you don't need to build a billion dollar business. We're told we have to because that's what TechCrunch reports on, and it's what YCombinator is there for.

[25:16] I also think there's another mode which is, what about lifestyle business? What about good cash flow positive lifestyle business and software that you serve a set of clients well. Where does that go? Does it have to be a billion? Does it have to be acquired or do you run it? Do you keep the software going? Do you expand? Is there cooperative models?

[25:38] We're starting to get to the point where there's enough tools in place that we'll see all sorts of things popping up, trying out these new models. That for me is pretty interesting.

Dan:  [25:47] It's amazing. An everyday person or someone who's not in tech can benefit from APIs, and a result could be that they can start a cash flow generating business using a low code/no code app and be able to benefit from all the history and technology and foundation that's been built by leaders like yourself.

Ross:  [26:02] Exactly. That's the whole point. If we don't allow more people to create in digital ways, then it's a very interesting societal question. We got rid of a lot of the mom and pop shops, thanks to the globalized supply chains, etc., where you have these big outlets. I'd love to see more of that come back. We might start to value that more post COVID.

[26:24] Another interesting mental exercise, there's a friend of mine I do it with every now and then, which is, what does the high street look like in 20 years? That's super interesting because I don't see a need to buy much from the high street, but maybe there's other experiences and other things that I'd go there for.

[26:41] I go to the high street for a social thing, not to get things purchased. I don't have an answer for this, but we make a lot of sense after about five whiskeys, and the next morning, we forget it all, but it's interesting.

[26:52] We're living in a world that will change pretty dramatically, and not always for the best either. We want to think about what things can we insert back into society that brings us back together again versus pull us apart.

Dan:  [27:05] I'd love to have you back on a round two to talk about high street, main street. The core of why we started the business is to help those businesses on main street transition to digital, and it does seem more likely that a lot of the legacy businesses that would have existed on main street may struggle to transform digital.

[27:23] However, there'll be a whole new breed of digital upstarts, like you're saying, that will be able to thrive, generate cash flow, and provide a great income stream for their families, and essentially live a happy life. Would [laughs] really love to have you in another conversation to decode high street.

Ross:  [27:37] All right. That'd be great. I'd love that actually. [laughs]

[27:39] [background music]

Dan:  [27:39] Amazing. Ross, thanks so much for joining us. This was great and take care.

Ross:  [27:44] Great. Thank you.

Dan:  [27:47] On the next episode of Decoding Digital.

Jim McKelvey:  [27:51] Occasionally, you can't copy something that hasn't been invented yet. What I want the world to understand is that all of us, if we're ready to do that little bit of innovation the two or three times in our life when it matters, the world's going to be a better place.

Dan:  [28:06] Co founder and Director of Square, Jim McKelvey.

[28:13] Thanks for listening to Decoding Digital. Make sure you never miss an episode by subscribing to the show in your favorite podcast player. To learn more, visit decodingdigital.com. Until next time.

[28:26] [silence] 

Transcription by CastingWords