Welcome to Words with Wynn! If this is your first time perusing my content and you’d like more of my weekly musings, subscribe below:
Robots are getting wild. We’ve got Boston Dynamics bots doing backflips, Tesla Optimuses DJ’ing launch parties, and don’t even get me started on the Thermonator..
Yet, despite all these advancements, robot integration into our daily lives has been slow to proliferate. Driverless cars are probably the most frequent manifestation of autonomy I see here in Austin on a daily basis, but little beyond that. Sure, we may eventually see an army of humanoids scurrying around cities solving our every whim, but that future appears to be a ways off. In the meantime, there are bathrooms to clean, floors to vacuum, and lawns to mow.
The stories in the press converge on the flashiest, sci-fi use cases (see: above), but it’s an open question as to what everyday autonomy will actually look like. Is it drone deliveries, autonomous skiffs, driverless cars or humanoid man-servants? Likely all of the above, eventually.
One framing of this debate is whether we see generalized humanoids or specialized alternatives. Billions have been poured into generalized humanoids, like Tesla’s Optimus, Apptronik’s Apollo, and Figure’s Figure 02, that promise to revolutionize everything– heavy industry, space exploration, mining, energy, you name it. This has produced quite a bit of hype.
But in the meantime, there’s an entire ecosystem of brass tacks roboticists being largely overlooked. Some of the coolest bots I’ve seen recently are being built by founders automating dirty, taxing jobs to drive a cleaner world, and I think there may be a market opportunity for a specialized platform player to emerge while the humanoid alternatives are still learning to handle a broom.
Let’s get our hands dirty.
Bathrooms, Carpets, and Lawns- Oh my!
Somatic
Autonomously cleaning a bathroom is a deceptive task. At first blush, if you’re like me, you’d imagine this routine chore to be relatively easy: scrub floor, clean toilet, rinse sink. It’s something that we humans (begrudgingly) handle with ease.
But then you consider the complexity required to actually automate such a task. You need your little cleanbot to open/close creaky stalls, unlock confusing door latches, navigate varying urinal and toilet configurations, spray different sink shapes/sizes/heights, mop pitched floors, and handle a hundred other minute details. All of which must be done at a speed, quality, and cost that’s competitive with today’s janitorial staff. It’s no small feat, but this is exactly what Somatic has done.
The YC-backed company is the brainchild of Michael Levy and Eugene Zasoba. The team have been partners in crime for over a decade, having previously scaled and sold edtech SaaS platform Chalkable in 2016. Michael has said that his scar tissue from the edtech space spurred him to look for precisely the opposite type of business next: no regulation, no competitors, and anything where tech giants like Apple, Google, and Facebook weren’t building. So, he started playing with robots.
His first foray into the space began with Trike, an autonomous wheeled delivery drone. It was a three year project that got the team building bots, and Michael points to this as their education in hardware. That work helped the team navigate the mind maze of where they could really find a market to win while providing them with a thorough understanding of what it takes to build bots. Ultimately, although package delivery is huge, it’s a crowded market and a highly variable problem set with constantly shifting routes, legions of competitors, and all manner of unexpected obstacles.
So, they turned to toilets.
Set piece tasks on consistent schedules within stagnant floorplans. Simple! Minus the myriad complexities listed above…
Even if you build the bot that can do it, working out the economics is a challenge in and of itself. The machine must be as quick and thorough as a human janitor for a fraction of the cost, all while being a seamless out-of-the-box solution for maintenance coordinators to implement. This gets to the heart of Somatic’s value proposition.
After five years of iteration, the startup has proven two major milestones:
Not only can its bots meet the cleanliness standards, but
They can be deployed in a seamless way
Their current machine operates using a robot arm and a number of attachments for its various functions: soap/water spray, dryer detachment, wet vac for floors, and a larger vacuum in the base for broader surface areas. In their most recent update, you can see the Somatic bot go from unplugging itself in the janitor’s closet through cleaning multiple bathrooms completely unaided. It’s taken time to nail commercial viability, but the team has continuously iterated with its customers and have now deployed almost a dozen robots, with ambitions to scale to hundreds in the next 24 months.
But high-touch, white glove pilot customers do not a scalable startup make. The second key to their solution is the actual deployment. The team has engineered their next generation bot to be delivered on-site to a customer, assembled by the local staff, and then programmed and deployed completely remotely. This is fundamental to scaling the Somatic fleet from a handful of machines to hundreds of customer sites in a labor-efficient way.
That said, quality of service and speed of deployment ultimately intersect at the question of cost. Somatic’s pricing is as follows:
A high level cost analysis can be run using Janitors & Cleaners stats from the Bureau of Labor Statistics. With the average annual wage of janitors around $36,250 (assuming average 40hr/week), Somatic’s 8hr weekday option is about 50% cheaper on an annualized basis. Of course, this comparison isn’t truly apples to apples— there are still a number of tasks that will require a human in the loop like refilling paper towels, flooded facilities, etc.
Nonetheless, at this price point, the Somatic bot is an attractive solution for janitorial service companies that face constant challenges with turnover, labor shortages, and quality consistency. I believe that the savings are significant enough to drive early ‘tech-curious’ facilities managers to pilot the Somatic bot, opening an opportunity for the startup to ‘wow’ them and expand their early champion base. Plus, having a Somatic machine on-site can also free up janitorial staff for other tasks like vacuuming...
Tailos
Austin-based Tailos is a home-town hero tackling industrial grade, commercial cleaning with a fleet of autonomous vacuum bots. To call it a Roomba on steroids trivializes what the team has built. Rosie, their machine, is effectively a miniaturized Waymo for all your vacuuming needs.
Full disclosure, the startup is a Capital Factory portfolio company, and we’ve deployed their bots at HQ.
Tailos was originally founded as Maidbot by Micah Green in 2015 during his undergrad at Cornell. A classic dorm room startup, the company was spawned from Micah’s experience working as a room attendant at the Statler Hotel on campus. This inspired him to tackle the labor-intensive problems that plague hospitality: bedding, vacuuming, and cleaning. While he initially envisioned a bot to change bedding, his customer discovery bore out that it was too early and capital intensive to develop a robotic solution to tackle that problem just yet (back in the mid teens). So, he doubled down on an equally pressing pain for hotels: vacuuming.
Micah teamed up with a Georgia Tech Robotics professor to scope the initial prototype. MVP in-hand, the team was able to secure their first angel checks and Micah dropped out of Cornell to go all-in. In the decade since, Rosie has come a long way. Although the initial Rosies could achieve the requisite cleanliness milestones, it took a number of iterations for the team to really nail the ease of use required to drive broader adoption. Micah’s quick to highlight that the average age of hospitality cleaning staff sits around 60 years old, so Tailos had to relentlessly simplify UX and refine its GTM. Before Rosies are adopted to work in concert with maids flipping rooms, the initial beachhead at any given property is often around public spaces like hallways, meeting rooms, and ballrooms, where property managers can gain comfort and confidence in the product while educating their staff.
Today, the team has nailed the customer pain point within hospitality, and has since expanded to other use cases like senior living, commercial real estate, and even cruise ships. The company’s fleet tops 1,000 robots deployed today across 12 countries, and their growth has been fueled by tens of millions in venture funding across multiple rounds. Rosie’s successes have even caught the eye of strategic players like Reckitt, owner of the brand Lysol, who led Tailos’s Series B. But why the interest, and how does a Rosie differ from a simple Roomba?
As you can imagine, the commercial demands for this use case are far more stringent than a simple consumer machine. Rosie isn’t much bigger than a standard Roomba, standing 12x16x6 inches and weighing approximately 18lb, but the platform packs far more of a punch. On the specs, the Rosie is about 10x faster than consumer robots, 5x more powerful, and has a much higher bar for durability (with most users running it 4-8 hours a day). These features are key for hospitality customers whose brands are built on tidiness. Additionally, the speed is fundamental to the ROI equation, where hotel property managers are often deploying Rosies to complement their maid staff in order to turn rooms more quickly (vacuuming while a human changes bedding, etc.) and drive higher throughput on their properties.
But where the business gets really interesting is the tech on top of vacuuming. Tailos machines are also outfitted with a host of sensors and supported by a software stack that further drives utility. On the sensor side, Rosies map where they’ve cleaned, edge cleaning percentages, and other data points like wifi strength, air quality, and mold levels across a property. This gives property managers direct insights into the cadence and cleanliness of their hotels at any given moment. The team has also built out swarm capabilities to coordinate multiple Rosies in real time for cleaning larger surface areas like ballrooms and corridors. This is all underpinned by an extremely simple UX that empowers customers to “press play, walk away”, dropping Rosie into any environment and letting it run autonomously straight out of the box.
But Is It Fundable?
So, the tech is viable and the solution is cool, but the question remains: is it venture-backable?
Beyond business model and capital intensity, a couple broad-based considerations come to mind: defensibility and market sizing.
For the former, there’s a small camp that might point to the hype around humanoids and consider their rapid iteration on a longer timeline. Can these specific use case solutions scale to a meaningful outcome before general purpose counterparts come down the cost curve? This risk factor is fairly easily addressed. Ultimately, I think our future cities will look like Disney’s Robots or Star Wars. Sure, we’ll have humanoid battle droids, but I think we’ll also have fleets of Gonk droids and repair bots scurrying around as well.
Today, general purpose humanoids are advancing exponentially, but they are still more novelty than tradesman. They will first be deployed at the high end for expensive use cases leveraging their unique strengths (3PL/warehousing solutions, manufacturing applications) before cost and scale make them a viable solution for menial tasks like everyday property management. Specialized solutions have a window in the meantime to gain scale in their own right while building out a broad-based platform to land-and-expand within their verticals. This is to say nothing of form factor, where the ideal machine to accomplish something like vacuuming probably isn’t a bipedal pushing a suction stick anyway.
For TAMs, startups like Somatic and Tailos are each going after sizable markets in their own right. But more than singular use cases, I think the real opportunity lies in building a true platform player. Following the Anduril playbook, the biggest winner in this space will build out a suite of machines, all empowered by a unified tech stack on the back end.
Once someone like Tailos owns a property’s vacuuming needs with a sticky product, it’s easy to sell through existing channels with an ever expanding product suite. This becomes a virtuous cycle as your portfolio of bots then coordinates as a network on your software stack.
I think that many investors and founders have come to a similar conclusion. Whoever can scale quickly enough within their chosen beachhead can compound their advantage. First movers will own sticky relationships that will be hard to displace. Once a property manager has an installed base of cleaning bots, there will be friction in switching costs. The leading service provider can then leverage its reliable cash flows to raise increasingly larger rounds to build a product suite in-house or through a roll up strategy, scooping up the subscale players in adjacencies to establish a dominant portfolio. All the while, the frontrunner will be accruing an increasingly differentiated data set on which to build their tech stack, wringing additional efficiencies across their install base as they gain scale.
Conclusion
Despite their lack of flash, these startups beyond the limelite might just be the winners in this robotics hype cycle. Humanoids will be huge… eventually. But the delta between now and viability, let-alone adoption, remains to be seen. All the while, specialist players are deploying in real-time, tackling real world use cases today.
Mountains of venture dollars will be torched in the pursuit of the T800, but it’s an open question which startup will be the Cyberdine and the surrounding hype has driven many of their valuations to far outpace the reality of the technology. Said another way, many techno optimists want shiny humanoids, but the winners for this vintage may just turn out be the Gronk droids along the way.
- 🍋