This post was written by our Co-Founder Alex Birch and originally published on medium.
It’s been a pretty momentous week for us at XY Sense.
After three years in R&D perfecting our sensor, some seriously gratifying pre-launch customer traction and a $5M seed round, we’ve officially come out of ‘stealth mode’ and launched our technology to the world.
But in preparing announcements and talking to journos, I realised I hadn’t ever taken the time to jot down why Luke Murray and I started this company and why we’ve spent so bloody long in stealth mode ensuring we got our solution just right.
So this is it. The XY Sense founding story.
The first thing to know is that XY Sense isn’t my first company. In the early 2000s, after building some custom applications to help corporates relocate into new buildings (without the need of thousands of printed floor plans manually marked up with highlighters), I cofounded workplace management software Serraview.
For years, I led the Serraview team here in Australia while we continued to extend our solution into a full suite of workplace management tools for facilities management, HR and real estate teams. We grew rapidly without having to raise any capital. Before we knew it, we had banks, telcos, and government departments as customers across Australia, the USA and EMEA.
By 2016, we were a team of around 80 people. One of that team was employee number 8, Luke Murray. Luke helped design and build much of Serraview’s technology. I’m also an engineer by trade and Luke and I would spend hours together diagnosing our customer’s problems.
One of those problems just kept rearing its ugly head.
That is, understanding how your employees are actually using your office space.
It was the single biggest problem we heard our customers talk about and with good reason.
Real estate is typically a company’s 2nd or 3rd biggest expense (behind people’s wages), and yet there was a complete data vacuum for these teams.
They couldn’t tell you with any real accuracy on a given day how many of their desks or meeting rooms were being used. They were having grads walk around with clipboards counting bags at desks and heads in meeting rooms but that only really told you how many people were in line of sight of a graduate on a Tuesday morning in April of last year.
Understanding desk or space use might seem boring for some but it ignited something in me. It might seem simple but it was a foundational problem for every company that no-one had solved.
Without accurate data on who’s onsite and working at desks and in meeting rooms, across a corporate portfolio it’s very difficult to manage so many people, teams, floors, and buildings. Do you really need that second office tower that’s going to cost hundreds of millions of dollars? What’s the best space design for the new building fit-out? Does marketing really need that extra floor?
These decisions have huge financial implications, requiring multiple stakeholders to work together but without accurate data there’s no common empirical ground. How do you know you’re making the right call?
Over-utilisation is just as much a problem as under-utilisation. A workplace is the beating heart of company culture, it exists to enable a workforce. It’s there to help foster collaboration, ideation, and delivery. Everyone hates never being able to find a meeting room or a desk near a colleague. It impacts you every day.
In that respect, the workplace data vacuum impacts every single employee of every company on earth.
It seemed crazy to Luke and I in the era of Google Analytics for your company website. There was no equivalent real-time data and analytics for your workplace.
Luke and I decided to step away from Serraview to solve this problem.
Navigating the idea maze
Having spent so much time with teams grappling with this problem, we knew that there were a few things that were non-negotiable.
We needed a technology that was affordable, flexible enough to understand utilisation of all spaces in an office environment (not just desks) and did not try to change people’s behaviour in order to operate. Having to log in or carry a device or tag could never work at scale.
Being able to show utilisation in real-time was critical. It was commonplace over the years to hear stories of teams hoarding space, and using excuses to justify why — from the “reserved for an upcoming project” that never actually started, to talking about contractors who weren’t on the network occupying space like ghosts. When you present a real estate strategy that is driven from utilisation data — if you can’t trust the data, then how do you know if it’s correct?
Another fascinating concept we learned about was that the “first shot is almost always right”. If you present data to the board of directors, it had better be right. If it isn’t then it won’t necessarily be corrected, even if it comes to light it’s wrong. Too many heads would roll. How do you build trust in utilisation data?
Simple. We had to make it so accurate that it would be good enough to show in real time, and then make that real time view available to everyone.
But how can you deliver that level of accuracy without compromising the privacy of the people in the space? No-one wants to be spied on. Technically ‘easy’ solutions like having video cameras sending a video stream off to the cloud would not only compromise our principles but would quickly become obsolete in a world where organisations valued privacy and information security above all else.
A unique solution to a sticky problem
With these guiding principles in mind, Luke and I spent the good part of a year investigating different technologies and settled on deep learning and computer vision. With that as a platform, we knew we could do far more than any of the fledgling occupancy sensors that had started to come onto the market.
We set out to build a hardware sensor that would capture space utilisation data in real-time but the only data that would leave the sensors would be anonymous X, Y coordinates of people. That’s why we call ourselves XY Sense.
This might sound simple. It wasn’t. At the early stages of deep learning prototyping, we actually had some experts laugh at us and tell us we’d “melt the chip” trying. It didn’t dissuade us.
With Blackbird Ventures backing us, we’ve built a pretty crazy team of ML engineers and experts who helped us achieve this goal. We have now designed a custom neural network that positions people onto a floorplan to a resolution of under 1m (mostly 30cm or less). We’ve squashed this algorithm down so it can fit onto a small low power device. And engineered a cloud solution that is serverless and can scale to consume billions of data points yet still display utilisation information on a floor plan view within 2 seconds of it actually happening.
Finally, knowing that the best utilisation data in the world is pointless if teams can’t easily access and consume it, we developed a scalable analytics platform that would allow us to simply present the utilisation information we were gathering in consumable dashboards and reports.
Pilots, here we come
After 2+ years of engineering, last year we commenced early customer pilots. Being engineers, we couldn’t help ourselves and actually managed the first installations to learn how to optimise the process and to ensure we were delivering the right results and experience to our clients.
Our traction has been humbling with pilots quickly converting to permanent building rollouts. And those same rollouts quickly expanded further to multi-building rollouts across the country. It’s hard to argue with an irrefutable real-time data feed. We’ve had CRE leaders explode in giddy disbelief as they walk around their boardroom and see an XY dot move across their floor plan. They recognised the impact this data could have when captured over time and applied to inform their long term workplace experience and real estate strategy. With XY Sense, their workplace data vacuum had become a thing of the past.
COVID-19 don’t kill our vibe
This year, we had grand plans for global expansion. Some of that is still happening but we’ve directed our efforts to adapt our solution to help teams safely return to work after COVID-19 lockdowns and better manage social distancing and space usage tracking for things like forensic cleaning.
For Luke and I, it’s clear that COVID-19 isn’t the end of the office. People are social creatures. They need to collaborate, and will focus work wherever they need to. Leaders need to impart knowledge to the next generation. Doing that over Zoom is nowhere near as productive as in person.
While we can speculate on time frames for the return to some kind of normalcy, the need to measure and understand how the workplace is being used is more crucial than ever. It will no longer be acceptable to not know who was onsite on a given day and where they sat. Spaces will need to change to support safe collaboration and teams will absolutely need to be able to book meeting rooms and desks on the days they are back in the office. We’re excited to be working with our clients to help to navigate this path out of COVID-19 and into the truly data-driven office of the future.
With our official launch this week, it’s been great to reflect on our journey so far and to remember our ‘why’.
We wouldn’t be here without our incredible team however, so I wanted to end with a shout out to them. Team, you’ve achieved what many said was impossible. You managed to turn this sticky problem and our vision into a scalable solution that is already delivering customer value. I can’t wait to see what we achieve together next.
Interested in learning more about what we’re building at XY Sense? Learn more about out our technology here.