The five-minute CIO: Dharmesh Shah, CTO, HubSpot

7 Apr 2017

HubSpot co-founder and CTO Dharmesh Shah. Image: HubSpot

“This is the first time ever in the history of software that we have a fighting chance at creating actually intuitive software,” said HubSpot’s Dharmesh Shah on the advent of AI and natural language processing UIs.

CTO Dharmesh Shah and CEO Brian Halligan founded HubSpot almost 11 years ago when they were both studying for MBAs at MIT.

HubSpot is a global inbound marketing and sales platform that is used by more than 18,000 businesses in 90 countries.

Last year, HubSpot announced plans to create 320 new jobs at its European headquarters at One Dockland Central, which will bring total employment at the company in Dublin to more than 600 people.

‘We have plans to grow from 300 to 600-plus people in Dublin over the next few years’
– DHARMESH SHAH

Prior to HubSpot, Shah founded Pyramid Digital Solutions, which was acquired by SunGard Data Systems in 2005.

Dharmesh holds a BS in computer science from the University of Alabama in Birmingham and an MS in technology management from MIT.

Named in the Inc Founders 40 list in 2016, he is an active member of the Boston entrepreneurial community, an angel investor in more than 60 start-ups and a frequent speaker on inbound marketing.

How did the idea of HubSpot come about?

Brian and I met at graduate school at MIT. He grew up in sales and I grew up in engineering. We had some shared passions, both in tech and enterprise companies particularly, but we both had a passion for small businesses. The reason for this was that we both previously worked in enterprise software in our earlier careers and the sales cycles were really long and painful, and the focus was on a small number of large companies that were paying a lot of money.

At the other end of the spectrum, you have consumer start-ups, and the problem is that they are very much like a lottery ticket, where there is a binary outcome: you either end up being the next Facebook or Google, or nothing at all – there is no in-between.

Brian went to work in a venture capital firm in the Boston area.We had a rapport and we discussed how we might actually start a company together some day.

During the year after Brian graduated, we were working on ideas and Brian was talking to venture-backed start-ups about how they were going about marketing – and they were struggling. On the flip side, I had started a blog and traffic was rising a lot more than any of these start-ups that had professional marketers and a budget.

There’s a thing called SEO, you can get traffic from Google, there are social media sites – and you need to go and get portfolio companies to do all these things. You can get WordPress for your blog, tools for your SEO.

To Brian, that all sounded like a science project. He pointed to all of these 10-person companies, and that’s where we had the idea of HubSpot.

All the changes in internet and marketing meant that most small and medium-sized businesses were struggling with doing this stuff and making that transition.

And that was the market opportunity. That’s where HubSpot was born.

What is your ethos on product building and execution?

The challenge we had early on was that there were good products and companies in each of the categories we found – in SEO, for example; social media tools, blogging tools, analytical tools etc.

All those things existed, but the problem we were out to solve was not a dearth of tools. The issue was getting them all to talk to each other and work together. It was not trivial.

And the reason we made that seemingly foolish choice was that we wanted to solve the problem the customer actually had. HubSpot wanted to create that all-in-one experience.

We were very inspired by Apple. Apple’s whole take is keeping everything integrated and all in one place. Apple is good at selling to people already using something, but it is unique at getting non-consumers to start consuming new things by simplifying products and, to do that, they control the entire ecosystem.

When you put all the pieces together, it is the best overall experience.

In theory, it sounds like a good thing and in practice, it worked out. But, from a product development perspective, there were never enough people on the engineering team to fulfil the vision.

Over time, we caught up.

How did Dublin help you solve your engineering challenges?

When we came to Dublin, we decided to make it our European headquarters. We were very deliberate in how we went about it.

I was super-fearful of disrupting our culture by not having everybody in the same place. We are a very culture-driven company. So what was bothering me was: how do we keep the magic when we transfer it over an ocean?

We took a leap of faith.

One of the things we did was atypical. Companies come to Ireland for various incentives such as taxes or a cheaper labour pool. We didn’t think about any of those things. What we did was we decided that in order for the culture to work, and for us to overcome the core fear that we had, was to make Dublin equal in every way to our HQ in Boston.

Culturally, first and foremost, we made it clear to staff: you are a HubSpotter and you have an inalienable right to everything. Even though it so happens that you are across an ocean, you will still experience the full force of it.

Our head of recruiting came over and was finding engineers as good as – if not better than – back in Boston and we said, ‘Let’s keep doing that’. And here we are, 300-plus people. It’s been a phenomenal story for us.

How has Dublin been from a tech talent perspective?

There’s a reason why we chose Dublin over London, and that was that the same advantage that Boston had. We had room to build an employer brand.

Boston is a small city versus San Francisco, and part of our motivation for Dublin was the same and that worked out well. It is a city where we can build an employer brand and a reputation as a great place to work – a lot of similarities to Boston.

We have plans to grow from 300 to 600-plus people in Dublin over the next few years.

It has gotten harder. The big issue we have now is less about talent and more about real estate and being able to grow. We are not the only ones to have discovered Dublin, evidently.

What kind of technical challenges do you have? Is your stack fully formed?

Yes, it is. We have all of those classic issues such as multiple dimensions, including scaling. The scale we have to deal with is not just the number of customers, but those customers themselves are growing in terms of the traffic they are getting to their websites, which was the whole reason for adopting HubSpot in the first place.

We have to deal with that geometric increase, but we have an amazing infrastructure-as-a-service team internally.

We are all on Amazon Web Services (AWS) and that helps us to scale economically.

One of the other challenges we have had over the last couple of years has been fixes around user experience and design. We went from having five or six full-time designers in the last two years to more than 24.

The whole mobile movement is a particular design and engineering challenge. You want to get to core features and business applications on iPhone or iPad or Android.

Our entire mobile team is here in Dublin. Overall, we have between 50 and 70 product engineering people here.

We are also at the beginner stages of understanding AI. This is a by-product of the machine learning revolution, and AI progress is seen across the industry, in terms of how chatbots and conversational UIs are becoming much more prevalent.

How do you see AI manifesting itself at HubSpot?

Since the first wave of AI, we have been working on it.

For our direct marketers, we have a chatbot product called GrowthBot and it works in Slack, Facebook Messenger and other platforms such as Google Analytics.

GrowthBot lets you type in a question like, ‘How was my traffic to the website last week?’, and it not only comes back with the answer, but lets you do competitive research; for example, ‘How much does Intercom spend on pay-per-click ads?’. And so, there are all these data sources out there.

We have all of this stuff in one bot. It is the number two ranked bot in the market on Slack’s app store.

We are piggybacking on AI, natural language processing and APIs in this area.

It is one of the big developments that have happened over the past year and a half, because all of the big tech companies such as Google, Facebook, Amazon and others have started to expose their machine learning and natural language developments as APIs.

We started this development a year ago, and it is getting thousands of users a week. It is super-popular on Slack and we are looking at Microsoft Teams as one of the other work-based collaboration platforms.

What predictions do you have for the near future of AI?

It has been fascinating to see. I am a big proponent of conversational user interfaces specifically, and the reason for that is that I have been in software for more than 25 years; and all of those 25 years, in the history of interfaces and how we interact with computers, it has been done on screens.

It started with desktop clients such as Windows and macOS. Then the browser came along, and it was still a desktop interface, but it was simpler and refined. Then the mobile interface came and now you can touch, swipe, pinch and zoom, and you are directly interacting with the data. Across all three generations, all of them have pixels in common.

And it works!

The problem with software developers is that we claim to build easy-to-use, intuitive software, but everybody says that. You still have to learn how to use the platforms to get the answers and outcomes you seek.

But that’s not necessarily intuitive.

You should just be able to just talk to the software and ask questions like, ‘How many customers do I have in Norway?’, and have it come back with the answer. That’s intuitive.

This is the first time ever in the history of software that we have a fighting chance at creating actually intuitive software. It is not going to replace desktop or mobile apps, but now we are going to get a whole new set of use cases that we weren’t able to do before, because we were constrained by screens and pixels and how much stuff you could design onto a screen.

Want stories like this and more direct to your inbox? Sign up for Tech Trends, Silicon Republic’s weekly digest of need-to-know tech news.

John Kennedy is a journalist who served as editor of Silicon Republic for 17 years

editorial@siliconrepublic.com