From Sidewalks to Software
$16 million. That’s what a16z and a group of co-investors decided to put behind Pit, a Stockholm-based AI startup founded by the same people who built Voi, the European electric scooter company. If that pairing sounds odd to you, you’re not alone. When I first saw this cross my feed, my immediate reaction was skepticism. Scooter logistics and enterprise AI automation feel about as related as a food truck and a Fortune 500 boardroom.
But here at agntbox.com, we don’t review vibes. We review tools. And the more I looked at what Pit is actually building, the more the founding team’s background started to make a strange kind of sense.
What Pit Actually Does
Pit is going after enterprise workflow automation, but with a specific angle that separates it from the crowded field of AI agents and no-code builders already fighting for IT budget. Rather than dropping a generic AI layer on top of existing software, Pit’s approach is to learn directly from the client how their business actually runs, then build custom software around that knowledge.
That’s a meaningful distinction. Most enterprise AI tools I’ve reviewed ask companies to adapt their workflows to fit the tool. Pit is flipping that. The product is supposed to come to you, study your operations, and generate something purpose-built. On paper, that’s a solid pitch. In practice, it raises a lot of questions I’ll get to in a moment.
Why the Voi Connection Matters More Than It Seems
Voi scaled across dozens of European cities, each with different regulations, different infrastructure, and different operational quirks. That’s not a scooter problem — that’s a workflow complexity problem. The founders spent years building systems that had to adapt to wildly different local conditions at speed. If you squint at it the right way, that’s actually decent preparation for building AI that learns how individual enterprises operate.
I’m not saying the scooter background is a credential. I’m saying it’s not the liability it looks like at first glance. The team knows what it means to deploy operational software in messy, real-world environments. That experience has value in the enterprise space, where the gap between a polished demo and a working deployment is where most AI tools quietly fall apart.
What I’d Want to Know Before Recommending It
As a toolkit reviewer, my job is to ask the questions the press release doesn’t answer. Here’s what I’m watching with Pit:
- How long does the learning phase take? If Pit’s product needs to study a client’s workflows before it can build anything useful, the onboarding timeline matters enormously. Enterprise buyers are not patient, and long ramp-up periods kill adoption.
- Who owns the custom software it generates? This is a real question in the AI-generated code space. If Pit builds something tailored to your business, what happens when you want to leave? Vendor lock-in through AI-generated proprietary code is a risk worth thinking about early.
- How does it handle edge cases? Every enterprise has weird legacy processes, exceptions, and workarounds baked into how they actually operate. The real test of any workflow automation tool is not the clean use case — it’s the messy one.
- What does the output actually look like? “Custom software” can mean a lot of things. Is this low-code tooling? Full application code? Workflow configurations? The specifics matter for evaluating whether your team can maintain what Pit builds.
Stockholm’s AI Moment Is Real
Pit is part of a broader pattern worth paying attention to. Stockholm has quietly become one of Europe’s most productive AI startup cities, and the combination of strong engineering talent, access to European enterprise clients, and now serious US venture backing from firms like a16z is creating a real cluster. Pit is not an outlier — it’s a data point in a trend that’s been building for a few years.
For readers who use agntbox.com to figure out what’s worth their time, my honest read on Pit right now is this: the concept is genuinely interesting, the funding is serious, and the founding team has more relevant experience than the headlines suggest. But $16 million in seed funding and a solid pitch deck are not the same thing as a proven tool.
I want to see Pit in the hands of real enterprise teams before I put it on any recommended list. The idea of AI that learns your business before building for it is exactly the kind of approach that could work — or could turn into an expensive, slow-moving consulting engagement dressed up in AI branding. The difference between those two outcomes is in the execution, and that’s what we’ll be watching.
đź•’ Published: