Building a Startup Is Not for the Faint of Heart: What a Decade of Osaro Can Teach the Rest of Us

Building a Startup Is Not for the Faint of Heart: What a Decade of OSARO Can Teach the Rest of Us

‍

Most robotics startups do not make it ten years. The math is brutal. The hype cycles are relentless. Customers and VCs are demanding and often value completely different things. And the technology you bet on in year one is almost never the technology the market wants in year five.

‍

I have been in robotics long enough to have watched companies I respected disappear, get acqui-hired into oblivion, or quietly pivot themselves into irrelevance. So when a robotics company crosses the ten-year mark and is still shipping, still scaling with real customers, still profitable in the ways that matter, that is worth paying attention to.

‍

I sat down with Derik Pridmore, CEO, and Gabe Zingaretti, COO, of OSARO, in their San Francisco showroom recently. I have known Derik since 2017. We were on opposite sides of deals back then, which means I know firsthand how hard they were to compete against. A decade in, they are still here, and they are winning quietly with customers most of us cannot even talk about by name.

‍

Here is what struck me from that conversation, and what I think every robotics founder, operator, and investor should be thinking about right now.

The shiny object will kill you

Every six months, robotics gets a new religion. Humanoids. LLMs in the loop. Vision language action models. Whatever the conference circuit is breathless about that quarter. There is real technology underneath some of it, and there is also a lot of theater.

‍

The pressure to chase comes from both sides. Customers do not always know what is possible, so they ask for everything. VCs read the same headlines you do, and they want to know how your roadmap maps to the latest narrative. The temptation, especially for first-time founders, is to nod along and quietly bend the company toward whatever buys you the next round.

‍

What OSARO has done well, and what I think most companies fail at, is staying ruthlessly focused on what actually works in production. They picked their bet early. Modular, flexible robotics. Intelligent piece picking. Tight integrations. High uptime. They have spent ten years deepening that bet rather than abandoning it every time the conference floor changed its mind.

‍

That sounds simple. It is the hardest thing in this industry to do.

Saying no is a strategy

Gabe said something during our conversation that I have been thinking about ever since. A lot of companies cannot say no. They get to 80 percent on a feature, the throughput is not quite there, and instead of grinding out the last 20 percent, they get distracted by the next promising thing. They do that three or four times and suddenly they have a graveyard of half built capabilities and no shipping product.

‍

Course deviating versus course correcting is the distinction that matters. Course correcting means you are still going north and you are learning as you go. Course deviating means you keep starting over. The companies that ship for ten years are the ones that course correct. The ones that do not survive are usually the ones that mistook one for the other.

‍

Saying no is unglamorous. It does not generate Twitter buzz. It does not put you on a panel at the conference. But it is what compounding looks like in robotics, and compounding is the only thing that gets you to year ten.

The hardest bets are the early ones

OSARO made a call in 2015 that piece picking was actually possible at production quality, and that the right way to do it was modular and cross-hardware. That bet looked aggressive at the time. Plenty of smart people thought it was too hard.

‍

Five years of technology development followed before they were really scaling. Five years. In a venture backed environment where the average partner is updating their model every quarter, five years of foundational work is an enormous act of conviction. It is also exactly the kind of bet that, when it pays off, gives you a moat nobody can catch up on quickly.

‍

I think about this every time a founder tells me they want to build something that already exists, but slightly better. The companies that win in robotics are the ones that picked something genuinely hard, committed to it before it was obvious, and then stayed committed long enough for the rest of the world to catch up.

Production is the only thing that counts

There is a line I keep coming back to. People work pretty well. They make mistakes, but they handle exceptions, they show up, they get the job done. To replace two people with a robot working eighteen hours a day, the algorithms are necessary but nowhere near sufficient. You need fleet management. You need integrations with auto baggers, barcode scanners, conveyors, AMRs, rack storage systems, and whatever else the customer already has on the floor. You need the system to keep running when the boxes have weird printing, when the SKUs change, when the customer wants to add a coupon dispenser next quarter.

‍

Most of the work that makes a robotics company real is invisible. It does not demo well. It does not make the keynote. But it is the difference between a pilot that quietly gets pulled and a deployment that scales to multiple sites and starts cloning itself.

‍

If you are a founder, the question I would ask is whether your roadmap is weighted toward demos or weighted toward production. Most of the companies that did not make it spent too much of their early years on the former.

Customers do not buy the Lamborghini

Gabe used an analogy I loved. If you want to get to work, you can drive a Prius. You do not need a Lamborghini. You could, sure, but the math does not work and you do not actually want to drive it every day.

‍

Robotics buyers are exactly the same. They are running a warehouse. They need to deliver toothpaste, or apparel, or pharmacy orders, fast and accurately. They are not buying the peak of the technology stack. They are buying ROI. The job of a good robotics company is partly education and partly translation, helping the customer separate what they want from what they need, and being honest enough to tell them when the want is going to break the ROI.

‍

The companies that can do that consistently build trust. The companies that just sell the Lamborghini eventually run out of customers who can afford one.

The contrarian bet on modular versus monolithic

The current narrative says one big general-purpose robot, probably a humanoid, will eventually do most warehouse work. OSARO’s bet is the opposite. The future they are building toward is robotic arms working in concert with AMRs, rack storage systems, and the rest of the warehouse, with every component doing what it is best at. 

‍

I am not going to tell you which bet wins long term. Nobody knows. But I will tell you that the modular bet is producing real ROI in real warehouses today, and the monolithic bet is mostly producing demo videos. If I were a customer with a brownfield to retrofit and a two-to-three-year build cycle for any new facility, I know which one I would be writing a check for this year.

What this all adds up to

Building a robotics startup is not for the faint of heart. The market is noisy, the technology is hard, the integrations are harder, and the patience required is roughly double what any reasonable person would sign up for going in.

‍

The companies that make it ten years are not the ones that chased the most trends. They are the ones that picked a hard problem, committed to it, said no to almost everything else, and ground out the unsexy production work year after year. That is what OSARO looks like up close. That is what the next generation of robotics winners is going to look like too.

‍

If you are a founder reading this, the question is not whether you have the vision. Most founders have plenty of vision. The question is whether you have the discipline to ignore the next shiny thing for ten years straight.

That is the bar.