- Saturday Startup Stories
- Posts
- You'll Never Mow a Lawn Again | Electric Shepe
You'll Never Mow a Lawn Again | Electric Shepe
Episode 33 of S³
Surely there isn't a startup (1) acquiring landscaping businesses, (2) to build a data engine, (3) to train embedded AI models, (4) so they can develop novel robots...
Yes, there is, and they’re called Electric Sheep.
TLDR Electric Sheep is making animal-like robots to tend the outdoors (lawn mowing, pruning, trimming…) by acquiring landscaping businesses to deploy their robots and build their own data engine to train custom AI models.
A unique data & business model
Electric Sheep’s robots are unique because they’re AI agents — and that’s not a future aspiration for them, that’s a reality today.
“All of the decisions are made by a neural network that are learning constantly. We’re inspired by how insects perceive the world, your ladybug, your caterpillar, how do they actually plan and do motion? We don’t want the 1,000 GPU model, we want a model that can run on fast embedded systems real time,” VP of Autonomy Mike Laskey explained while a few Ram robots striped the lawn of their HQ.
AI-based robots are incredibly dynamic. For example, a few minutes after I showed up to film this episode, I asked the team if they could run a few Rams outside for filming. The guys shrugged, agreed, then placed some Rams on a nearby lawn, turned them on, and watched as the robots immediately got to work. No preprogramming, no local navigation information, no human direction — just a stereo camera connected to Electric Sheep’s embedded, physically based model.
To have AI work well at a massive scale you need a lot of training data, that’s where Electric Sheep’s business model shines. Rather than generating synthetic data for training, relying on partner data companies, or solely utilizing open-source data Electric Sheep is collecting their own. They’ve started by acquiring landscaping companies, deploying their robots on-site to improve margins, all while generating a waterfall of real-use data.
Creating Wall-E
Electric Sheep decided to use their S³ feature to announce their second robot named Verdie. I’m sort of a big fan.
Verdie is a self balancing, tool-using robot that is being trained to navigate and tend the outdoors. Verdie is a nice little guy, it’s hard not to like him — his fun eye animations, body design, and the way he calmly navigates around you leave the impression of a thoughtful and caring creature, more like an animal than a robot.
That’s not by mistake, this is an intentional design philosophy Electric Sheep has.
“There’s also a philosophy to this, we don’t think robots are going to be these really complicated humanoid machines. We think robots will be like animals, sort of like how we domesticated animals and made them really helpful,” CEO and Cofounder Nag Murty explains.
Making Wall-E Real
Most of the companies we feature on S³ are pretty early stage in their journey, as a result, it can be hard to imagine how some of these companies could ever become as big as Amazon, Google, Tesla, and Meta despite having similar origins. However, once in awhile, you can see a glimmer of how monstrous growth like that might occur.
For example, it’s pretty daunting to look at Tesla or Waymo and imagine competing with them on self-driving from scratch. They’ve been working at it for years and in Tesla’s case they’ve collected millions of miles of driver data from their fleet of vehicles. How could you ever compete with a data and training moat-like that?
Then out of nowhere, actually out of the garage of a wedding venue in San Jose, a focused and dedicated group of engineers begin acquiring landscaping companies and deploying their own robots across these businesses to build a real-use data engine.
If they’re not already, I expect Electric Sheep’s models will soon be the best in the world at understanding off-road outdoor navigation. Companies like Electric Sheep remind me that there are plenty of great ideas left to build if only we follow our curiosity, and want to mow our lawn of course.
Keep on building the future,
— Jason