The Mage Pivot: How Customer Pain Points Transformed Their ML Platform into a Data Pipeline Solution
Sometimes the most valuable pivot isn’t about discovering a new market – it’s about solving a more fundamental problem in your existing one. In a recent episode of Category Visionaries, Tommy Dang shared how Mage’s journey from machine learning platform to data pipeline solution revealed an essential truth about building foundational technology.
The Original Vision
Drawing from his experience at Airbnb, Tommy saw an opportunity to democratize machine learning tools. “We had disparate tools for pulling data, extracting data, transforming data, storing the data, building training sets, reusable data sets, feature engineering, also deploying models, training machine learning models,” he explains. The initial goal was clear: make these tools more accessible to a broader technical audience.
Early Success Masking Deeper Needs
Mage launched their cloud-hosted ML platform in early 2022, targeting early-stage companies from seed to Series C. On the surface, things looked promising – they had paying customers and thousands of sign-ups. But beneath these positive signals, they discovered a more fundamental challenge.
The Pivot Point
“Although there’s an appetite for machine learning, a lot of these companies, they want to use machine learning, everybody wants to use machine learning,” Tommy notes. “But what we found is they actually struggled with a more urgent data challenge early in their journey. And it is just the movement of data, the transformation of data, the integration of data.”
This insight led to a crucial decision. Rather than pushing forward with their original vision, they decided to focus on solving this more fundamental need. They took one of their core internal technologies – their data pipeline component – and spent several months preparing it for open source release.
Validation Through Open Source
The response validated their hypothesis dramatically. “We open sourced that earlier this year, and it’s been catching on like wildfire,” Tommy shares. The metrics backed this up: over 2,000 GitHub stars, 400-500 Slack community members, and numerous companies putting it into production.
Finding Their Edge
The pivot allowed Mage to differentiate themselves in the data pipeline space through four key advantages:
- Superior developer experience
- Built-in engineering best practices
- Data-first design
- Simplified scaling
As Tommy explains, “As many people that use Airflow, I haven’t met one person that said they love working in Airflow.” This insight helped them focus on making their tool not just functional, but enjoyable to use.
The Category Question
Rather than trying to create a new category, Mage positioned themselves against a known quantity. “We position ourselves against Airflow. So we are a modern replacement for Airflow… I think that’s just the easiest way for data engineers to understand what we do,” Tommy explains.
However, they’ve maintained a distinct identity by focusing specifically on data pipelines rather than general orchestration. “We are much more specific than that. We say we are a data pipeline tool for transforming and integrating data.”
Looking Forward
The pivot has shaped Mage’s long-term vision. “We see a world where Mage is the go-to data tool for early stage companies, mid sized companies. It’s a tool that you think about and you spin up as soon as you start a company, as soon as you have any database set up.”
For B2B founders, Mage’s journey offers a valuable lesson about product evolution. Sometimes the path to success isn’t about pushing harder on your initial vision, but about listening carefully to discover what foundational problems your customers need to solve first. By focusing on these fundamental needs, you might find an even bigger opportunity than you originally imagined.