It’s a new chapter for our team and a natural next step in a journey that started years earlier, long before TensorStax was even an idea.
Like many of us on the TensorStax team, I was drawn to the idea that software could reason, take action, and operate on its own. But across different roles and companies, we kept running into the same issue: no matter how advanced the models were, data pipelines were always the bottleneck.
As we worked closely with data teams, that gap became impossible to ignore. The hardest part wasn’t training models or generating output. It was cleaning, mapping, and verifying data across messy, real-world stacks. This was a shared frustration, and it became the starting point for what we would build together.
Over a weekend, our team put together a simple CLI in early 2024. You’d point it at a dataset, it would inspect the structure, write code to clean it, and push it back. It was meant as an internal tool, a quick way to unblock ourselves, not a product.
Then we showed it to a customer.
Their reaction made it obvious: this was the real pain point.
That weekend project became TensorStax, founded in late 2023 with a clear thesis: large language models could generate solutions, but without verification and deep integration into existing tools, they couldn’t be trusted to run real data systems.
As we talked to more customers, a pattern emerged. Teams were running Airflow, dbt, Snowflake, and other tools side by side, and they wanted systems that could reason across all of it.
We rebuilt TensorStax around that idea: autonomous AI for data engineering that could build pipelines, verify them programmatically, and adapt as things changed. That focus led us to raise a pre-seed in late 2023 and a seed round in late 2024.
Along the way, we spoke with several great companies about partnerships and acquisitions. What ultimately made Snowflake stand out was alignment on vision, culture, and where data and AI are headed.
Snowflake isn’t bolting AI onto an existing stack. The company is re-architecting the data platform for a world of agentic systems that run natively, securely, and at scale. And culturally, there’s a refreshing sense of urgency: a startup mindset focused on building with conviction, moving fast with discipline, and leading the next generation of enterprise AI rather than chasing it.
That vision matched exactly what we had been building.
TensorStax is now part of Snowflake.
The specialized tooling we built to make agents effective at data engineering now lives on inside Snowflake, including Cortex Code, which launched this week. It’s designed to help developers build the next generation of data and AI applications using systems that don’t just generate output, but reason, verify, and operate autonomously.
I’m incredibly grateful to our customers, investors, and early believers who supported us from the very beginning. We’re excited to keep building as part of Snowflake and join them in their mission to help every enterprise unlock its full potential through data and AI.
.avif)
.avif)
