Kohsuke Kawaguchi + Launchable = Smarter Testing, Faster DevOps
How Kohsuke's experiences and challenges in the DevOps industry lead to creating Launchable.
Key Takeaways
As the creator of Jenkins, I had the unique privilege to go see many software development teams around the world. And a consistent theme has been emerging: relying on data to drive velocity and quality.
Focusing on harnessing data, Launchable is using the power of machine learning to empower engineering teams to speed up bloated testing cycles with intelligent tooling.
Software development, meet big data.
That’s why Harpreet and I are starting Launchable. As the creator of Jenkins, I had the unique privilege to go see many software development teams around the world. And a consistent theme has been emerging.
Data is making a profound impact to all kinds of businesses, where gut and instinct-driven decisions are replaced by quantifiable metrics and models, which enables a whole new angle of process/efficiency improvements. Yet in the software development field, we still largely seem to be driven by our individual experiences, beliefs, and instincts. Quite simply, this IS the obstacle to the launch of your new software, hence the name “Launchable.”
Using Data for Smart Testing
Where this is really evident is testing. In so many projects, every pull request goes through a CI build that takes an hour or more. Then somebody comes in and does a code review, resulting in a minor touch-up that requires another whole hour of testing.
In other projects, so many tests are amassed and so many engineers are working on it such that it’s simply impractical to run every test against every change. So people rely on arbitrary subsetting of tests (think “smoke tests”) but those tests are chosen completely by instinct. And other tests are run on a fixed frequency (think “nightly end-to-end tests”).
I’ve been working on solving this problem by applying machine learning. At Launchable, we think we can predict with meaningful accuracy what tests are more likely to catch a regression given what has changed, and that translates to faster feedback to developers.
What if you only need to wait for 5 minutes instead of one hour to get your colleague to review your code because we can choose the right 10% of the tests that give you 80% confidence?
What if instead of running 6hr nightly end-to-end tests that tests everything that changed in the past 24 hours, you run an adaptive 2hr subset of end-t-end tests 3 times a day, and get feedback within the same day?
If any of this sounds interesting to you, sign up to our beta invitation list!
As exciting as this is, there’s so much more that we can do to drive more efficiency in the software development process. We’ll talk more about that in the coming days. Follow us on Twitter, LinkedIn, Facebook, or our mailing list.
Another thing that excites me in this new chapter of mine is that I get to build a company and a business from scratch again. In 9 years that I spent with CloudBees, I learned a lot about how there’s so much more to building a successful startup beyond technology and engineering.
This time around, as co-CEO, I intend to put that learning into practice. I’m on this journey with Harpreet Singh, who has been my partner in crime for what, maybe 15 years? Between the two of us, we have more than 25 years of combined experience in this space, we built a successful Jenkins business, and we learned to work together flawlessly. To me, working with people I respect and I enjoy counts a lot, and I believe a great team is a foundation that leads to a great outcome.
Launchable is hiring.
If you are passionate about software developer productivity and interested in working with experienced entrepreneurs well known in the field who cares about the team, I think we have a unique value to offer. You know how to reach me.
