How Claro Pivoted, Pre-Sold, and Landed Their First Enterprise Client in 3 Months

There’s a special kind of wisdom that comes from getting it wrong, fast. For Matteo Fava, co-founder of Claro, the early days of startup life were a lesson in that and in letting go.

After a decade leading global data teams at Delivery Hero and seeing firsthand how messy, fragmented business data could derail even the most promising AI initiatives, Matteo decided to solve it.

“I’ve seen every data problem you can imagine, and it was always the same story: data’s there, but it’s unusable.”

Together with co-founder Tameesh Biswas, a former engineer at Yelp and Autodesk, Matteo launched Claro in Berlin. The goal? To help companies finally make sense of their data so they could build AI products without months of prep, labeling, or painful model iteration.

But the first version of Claro didn’t stick.

“We built too early, too fast. We had conviction from talking to a few people, and thought that was enough. But when we tried to sell it, we realized this wasn’t urgent enough for companies.”

The Pivot That Changed Everything

Instead of powering through, Matteo did something harder: he stopped. The team paused all platform development, scrapped what they’d built, and went back to square one.

“We stopped coding and just started building customized Python notebooks for each company we spoke to. No interface, no product. Just solving their problem manually.”

That back-to-basics approach led to a new opportunity: companies with enormous inventories, sometimes half a billion items or more, that needed to classify and enrich their data across thousands of categories.

“We had a client that needed categorization across 12,000 categories. Imagine writing prompts for that. That’s when we knew this wasn’t just a problem. It was the problem.”

Claro began building tailored AI models for each customer, designed to work with their unique, often chaotic datasets. LLMs weren’t accurate enough at this scale, so they leaned into small language models (SLMs), knowledge distillation, and model compression.

Selling Before There’s a Product

When asked what milestone he’s proudest of, Matteo doesn’t hesitate.

“Selling before we were ready to sell. That was the hardest part. We used Streamlit demos and scraped together prototypes just to get people to see the value. It worked.”

Their earliest customers didn’t even touch an interface. They dropped data into a bucket, Claro processed it via API, and they got back clean, enriched, categorized results.

“From pivot to paying customers, it took three months. Then one month later, we closed our first enterprise client.”

Growth Through Founder-Led Discovery

Claro’s go-to-market strategy was raw, founder-led, and deeply manual by design.

“I’m not a salesperson. I just started talking to people in my network, offering my expertise. I’d ask about their data challenges and see if I could help. That was it.”

In classic growth hacker fashion, Matteo even flipped cold outreach on its head.

“If someone spammed me with a tool, I’d reply and pitch Claro. Like, ‘Hey, your product categorization looks off, want to chat?’”

It wasn’t scalable, but it was real. Those conversations didn’t just help him close deals, they helped him sharpen Claro’s messaging, validate pain points, and decide which features were worth building.

The Funding Came After the Focus

What’s wild is that all of this, the pivot, the prototype, the traction, happened mid-fundraising.

“Some VCs saw one version of Claro, then two weeks later, a totally different one. We even went from three founders to two during that time.”

But the clarity paid off. Claro closed a €650K pre-seed round backed by Atlas SGR, Antler, and Founders Factory, among others. The funding is fueling product development, onboarding automation, and building a smoother self-serve experience.

Balancing Product Vision with Customer Reality

Even now, the biggest challenge isn’t technical. It’s staying focused.

“Every enterprise wants something slightly different. It’s hard not to become a consultancy. We constantly ask ourselves Is this a real feature or a one-off request?”

Matteo is clear: Claro isn’t about chasing vanity metrics or collecting logos. It’s about solving a massive, overlooked bottleneck in AI development and doing it right.

“We don’t want to scale too fast. If we onboard too many clients too early, we lose the ability to learn. And that learning is everything right now.”

Advice for Founders: Kill the Thing That Isn’t Working

When asked what advice he’d give to aspiring founders, Matteo doesn’t talk about tools, frameworks, or pitch decks.

“Focus. And don’t be afraid to kill what you’ve built. That’s the hardest part. But if no one wants it, if you’re chasing people just to get feedback, that’s your answer.”

“Talk to the same people multiple times. That’s how you learn if the problem is really urgent. And if it’s not? Move on.”

Looking back, Matteo has no regrets about that early product they had to scrap. If anything, he sees it as a necessary rite of passage.

“It’s funny. A year later, almost all the competitors from our first pitch deck have pivoted, too. We just got there six months earlier.”

The Claro Mission

Claro isn’t trying to be another flashy AI startup. It’s trying to be the useful one. The one that finally helps teams turn their messy business data into something actionable, automatable, and scalable.

“We want to make advanced AI accessible for non-experts. That’s what excites me, giving teams the tools to build without needing an ML department.”

And if you’re in the early stages of building something yourself, Matteo’s got one final note:

“You don’t need it to be perfect. You just need to do the thing. Talk to people. Show them something real. And be willing to let go when it’s not working.”