Berlin startup Tower just pulled in €5.5 million to solve a problem every data engineer knows too well. AI can write code faster than ever, but getting that code to actually work in production? That is still painfully hard. Tower wants to close that gap, and some of the biggest names in data infrastructure are betting it can.
What Tower Does and Why It Matters Now
1 The gap between what an AI coding assistant can produce in minutes and what a production system actually needs to stay alive is the problem Tower is trying to close. 3 Founded in 2024 by former Snowflake engineers Serhii Sokolenko (CEO) and Brad Heller (CTO), Tower provides a single platform where humans and AI agents can work together to take AI-generated code and turn it into software systems that businesses can rely on.
Think of it this way. Tools like GitHub Copilot, Cursor, and Claude Code help engineers write data pipelines in minutes. But writing code is only half the battle. Testing it, fixing issues, deploying it, and keeping it running on real company data is where things break down.
12 “It’s easier than ever to write functional code, but it’s still difficult for humans, and even more difficult for AI agents, to test it, fix issues, deliver it to production, and operate it. That’s what we’re here to fix with Tower,” said Brad Heller, CTO of Tower.
This is what Tower calls the “last mile” problem. And it is only getting worse as AI coding tools get faster.
Tower Berlin startup AI data engineering production platform funding
Who Backed Tower and Why That Matters
1 The startup has raised €5.5 million (approximately $6.4 million) across a pre-seed and seed round, backed by investors including Speedinvest and DIG Ventures.
But it is the angel investors that tell the real story.
4 Additional investors include Flyer One Ventures, Roosh Ventures, Celero Ventures, and Angel Invest, with participation from angels including Motherduck’s CEO Jordan Tigani, Datadog’s CEO Olivier Pomel, Harvey.ai’s VP of Engineering Ben Liebald, and Taktile’s CEO Maik Taro Wehmeyer.
These are not random checks from passive investors. These are builders who run some of the most respected data and AI companies in the world today. When the CEO of Datadog and the CEO of Motherduck both put personal money into a seed stage startup, it sends a clear signal to the market.
“AI has made it easier to write data pipelines, but getting them to run properly in production is still hard, and only getting harder. Serhii and Brad have lived this problem first-hand.”
12 Melissa Klinger, Partner at DIG Ventures
1 Speedinvest’s Florian Obst, who led the firm’s investment, pointed to the multi-tenant architecture as a key differentiator: a platform designed from the start for fast integration and rapid iteration, rather than retrofitted from an enterprise monolith.
The Founders Saw This Problem Coming at Snowflake
Sokolenko and Heller did not stumble into this idea by accident.
11 Before founding Tower, Sokolenko worked at Databricks in the Unity Catalog compute group, at Snowflake on a high-concurrency, low-latency version of the data warehouse, and at Google Cloud Dataflow, building a unified batch and streaming data platform. 15 Tower is startup number three for both Brad and Serhii. 16 Sokolenko previously co-founded two other product startups in text analytics and legal tech, along with two IT consulting companies in BI and health tech. 14 Sokolenko shared, “After meeting at Snowflake, Brad and I started Tower in late 2024 because we saw a major shift happening in data engineering. For years, the field had been shaped by overly complex big data platforms. But suddenly, a new generation of data engineers was emerging, building with open source tools and writing data applications directly in Python.”
The infrastructure around those engineers simply had not kept up. That realization became the foundation for Tower.
How Tower Works Under the Hood
3 Designed for teams shipping scalable data and AI products, the platform unifies Python-native orchestration, managed Iceberg storage, and control plane APIs into a single multi-tenant system. This eliminates the need to stitch together disparate tools.
Here is what makes the technical approach stand out:
- Open data format: 1Iceberg has become the de facto open standard for analytical storage, compatible with Snowflake, Databricks, and most major data engine vendors. Tower customers are never locked into a single stack.
- Fresh data for AI: 3This makes customers the owners of their data and ensures AI agents receive fresh, company-specific information for accurate decisions and avoiding hallucinations.
- Single environment: 14Tower combines analytics, storage, and processing with AI-human collaboration.
- Cost-effective architecture: 14Sokolenko stated, “Tower is easier to use, better integrated with the other tools developers have in their toolbox, and works natively with coding agents. Because Tower’s architecture is dramatically simpler and built on open technologies, it’s also substantially more cost-effective.”
1 Gaurav Saxena, director of engineering at Ford Motor Company, offered a customer-side view. Apache Iceberg, he said, represents genuine strategic value for enterprises, but “operating it effectively demands skills and ongoing maintenance that many data teams aren’t staffed for. What’s compelling about platforms like Tower is their ability to remove that operational overhead.”
Early Traction and the Road Ahead
The numbers so far suggest Tower is not just an idea on paper.
3 **In February 2026, just a few months after launch, Tower exceeded 200,000 runs of 30,000+ unique apps, and its Python SDK reached 70,000 monthly downloads.** 3 Builders worldwide are adopting Tower as the “last mile” platform of choice, particularly among those developing Vertical AI services and SaaS. 14 Currently, the team has over 12 nationalities, including team members from Greece, Turkey, Sri Lanka, the US, Canada, and the UK.
The timing for Tower could not be better. 36Around 41% of all code written in 2025 was AI-generated, and current trajectories suggest that number will cross 50% by late 2026 in organizations with high AI adoption. 35Gartner predicts 75% of enterprise software engineers will use AI code assistants by 2028, up from less than 10% in early 2023.
Every line of AI-generated code still needs a home where it can actually run. That is exactly where Tower wants to live.
1 The market it is entering is competitive. Snowflake, Databricks, and a wave of newer data infrastructure startups are all investing heavily in the same AI-era data engineering story. 1 What Tower is betting is that none of them are focused specifically on the problem that emerges after the AI finishes writing the code. 2 Tower plans to use the new funding to grow its go-to-market team and enhance the capabilities of its platform.
As AI keeps writing more code at record speed, the bottleneck is no longer creation. It is production. Tower is building the bridge between what AI dreams up and what businesses can actually depend on. For data engineers everywhere who have felt the frustration of pipelines that work in testing but crumble in the real world, this startup might be worth watching closely.
What do you think about Tower’s approach to the “last mile” problem in data engineering? Drop your thoughts in the comments below.
