Before It Clicked From Whiteboard to $1.5B: Airbyte's Origin Story
When Michel Tricot left his job in 2019, he didn’t have an idea yet — he just knew he wanted to start a company.
Over the next year, he and his cofounder cycled through ideas: remote-controlling self-driving cars, fintech for gig workers, a healthcare data-sharing platform, and a marketing analytics product that actually had paying customers… until COVID exposed it as a “nice to have,” not a must-have.
Instead of forcing mediocre product–market fit, they did the hard thing: admitted they had the right team, wrong idea, and went back to the whiteboard. From there, the team ran weekly Zoom sprints, ranking ideas by founder–market fit, team excitement, and how painful the problem sounded in real interviews.
The pattern that kept resurfacing: everyone was struggling to move data reliably between tools. That insight became Airbyte — an open-source way to build and share data connectors, now used by thousands of companies and valued at around $1.5B.
I’m Sunny Rekhi, host of Before It Clicked. In this episode, Michel walks me through:
The dead-end ideas before Airbyte
How they ran structured weekly idea sprints
The moment he realized “we have the right team, wrong idea”
How the open-source Airbyte thesis finally clicked
Why agents are the next big consumer of data
Chapters
(00:00) Intro & where Airbyte is today
(05:07) How Michel thinks about starting a company & idea validation
(07:22) Pivot #1: remote-controlling self-driving cars
(10:44) Pivot #2: fintech for gig workers (the YC idea)
(16:06) Pivot #3: healthcare data-sharing & discovering “data movement”
(21:22) Dexterity/DataLine: marketing analytics & the COVID “vitamin vs painkiller” moment
(27:18) Realizing it’s the wrong idea & deciding to pivot again
(30:40) Weekly Zoom whiteboard sprints & how they ranked ideas
(40:17) Building an open-source Segment prototype → the Airbyte insight
(46:50) Airbyte v1, design partners & early traction
(55:25) Community explosion, fundraising & Airbyte’s future with AI agents