We're excited to announce that Tower is becoming part of MotherDuck!
Tower's engine already powers MotherDuck Flights, their AI-native data pipelines product. By joining forces, we're making secure, production-grade execution of agent-generated code a native capability of MotherDuck. We want agents to move beyond answering questions. Together, we'll enable agents to build and run the pipelines, apps, and systems that act on the answers directly in MotherDuck.
What we built and why
We started Tower to solve a specific need: If you are a data engineer, writing Python was never the hard part of the job. The hard part was everything that came after it. Packaging it into something portable, handing it credentials without leaking them to someone else, isolating it from whatever ran before, after, or around it, putting it on a schedule, and knowing what actually happened when it changed or broke. Many teams had rebuilt versions of that infrastructure themselves, often on heavier platforms than they needed.
So we built a new platform. It's Python-native, serverless, lightweight, and scales elastically to meet the demands of tiny startups and large enterprises. It's aligned with our vision of the future of data engineering, which is built upon open tooling and open data formats. The exact same Python code can move from local development to production in Tower with nothing to think about in between.
And all of that has held up. It held up well enough that when MotherDuck needed an engine to power Flights, Tower supported the rapid launch of Flights.
Everyone thinks the bottleneck moved
Models got good (really good) at writing code faster than most people expected. But it turns out that agents struggle with the same problems that engineers do after they've generated some Python code: Where does it execute, under what context, using which resources, and how do the humans and the other agents know what happened?
Tower already spent years building exactly that for humans and, almost serendipitously, it aligned perfectly with the agentic world. We've made it our mission to make every agent a data agent and the market moved toward that position rather than away from it.
Today, Tower runs over a million Python jobs every month and more than 90% of those apps are generated and maintained by agents.
Tower and MotherDuck: Two products moving in the same direction
MotherDuck first came to us as a customer for Flights and became our largest almost immediately. They had already built a data warehouse optimized for agentic workflows and they knew how important it was for agents to be able to build and maintain the pipelines that integrate data into their warehouse automatically.
But working with them never felt like a normal vendor/customer relationship. For months, our engineers designed and shipped the product together, and the most important work kept landing somewhere between the systems. Eventually, maintaining that separation made less sense than just becoming one team.
The decision made sense for Tower because of the strong cultural alignment. From an engineering perspective, both teams are strongly engineering-first, skeptical of the traditional big data approach, committed to open foundations, and willing to obsess over the gory details of building production systems designed for operational trustworthiness.
As founders in the data space, we share a vision for where the data ecosystem is headed with Jordan Tigani, MotherDuck's CEO, and the rest of his leadership team. This became quickly evident early on while working together. It is clear to all of us that we were building toward the same future in much the same way.
What this makes possible
Tower and MotherDuck are better together. MotherDuck gives agents and people the data, context, and the fast, isolated compute they need to understand what's happening. Tower gives the work they create a durable, isolated place, designed for secure, production-grade execution, to deploy, schedule, scale, and monitor pipelines, notebooks, scripts, and apps. Together, they close the loop from raw data to answers to action, whether the work begins with a prompt or in a text editor.
Bringing together Tower and MotherDuck gives agents (and humans) a single, fully-integrated platform for automatically integrating data and answering questions across your entire data estate.
What this means if you use Tower
- For existing paying Tower customers, your service will continue to work uninterrupted and we'll make sure there's a smooth transition from Tower to MotherDuck Flights.
- All free Tower accounts can sign up for a free MotherDuck account and get started today!
Want to get started? Speak with a MotherDuck product expert to learn more about how MotherDuck Flights closes the AI powered data loop, or sign up for a MotherDuck account to explore the agentic data stack powering real-time queries.