Summer 2019. Bright sunny day - birds tweeting about. Soothing sound of water crashing in the water fountain in my backyard. In short - a perfect day.
Kohsuke and me met in the sunroom of my house for a conversation.
We both weren’t happy.
While, we maybe like chalk and cheese in looks (he is tall 6' 5 inches & I am short - 5' 7 inches did you really have to ask - how rude!), after a dozen years of working together, we can complete each others sentences. But this time, we did better. We uttered the same sentence…
“The state of software delivery isn’t moving fast enough & the answer is not crappy software thrown over the wall at a faster pace!”
How did we feel this especially after we spent a dozen or so years helping companies move faster?
Kohsuke is the creator of Jenkins, arguably the most important DevOps CICD product in the market. While I was the business guy behind Jenkins. Together, we helped build CloudBees (recently valued at a $1B) into an organization to help bring DevOps en masse.
“Did we really need to do something about the state of software delivery? If so what?”
I had a great job as Head of Product at Atlassian Bitbucket, Kohsuke meanwhile as the CTO of CloudBees & creator of Jenkins has sealed his legacy.
Should we step out of our comfort zone and build out another startup? And why?
The answer was a resounding Yes! The why behind the yes turned into our mission.
Why start Launchable? To make software delivery data-driven!
Our motivation was simple. We owe everything in our lives to being developers. We both wanted to contribute & give back something to the software development community. Build something of significance so that we can look back & say “hey we helped people meaningfully” was a mission that we were both passionate about.
We were also passionate about the opportunity to make software delivery data-driven!
We talked about how every significant human achievement from here on would have software underpinnings. We talked about being frustrated. Frustrated, that in an increasingly data-driven world, software delivery and development has remained a mix of art & science. Somehow sales & marketing has taken a march over engineers in being data-driven.
We wanted to take our software team community and help them evolve to being data-driven scientists - using data and ML to drive software delivery. Data-Driven DevOps or ML Driven DevOps is what we called it.
How does Launchable help software teams? We are a force multiplier for software delivery.
Developers spend up to 36 hours per week on bad code & a productivity loss of 9.25%. The global GDP loss is a staggering $85B/year dollars (2018 numbers - Stripe report).
Those are some big abstract numbers. Let’s humanize them.
In real human terms, the numbers hide the human story. The human story is of overwork, anxiety, stress and relentless pressure with the inability to do something about it.
A team lead with one of our customers (& more importantly a mother of two) in Germany mentioned that her team was frustrated by the slow software delivery cycles. The productivity loss was more like 30%. Imagine - a third of the day wasted - a third of the day of your most expensive resource flushed down the drain. She stays up after dinner, after putting her kids to sleep, waiting to see if her build has gone green. She sees her DevOps team working over time to help but not enough help is coming their way too.
This is not how software delivery should be! And. We want to can fix it.
As we looked at the software delivery cycle - we came to the conclusion that just building software delivery pipelines is NOT enough.
The delivery pipeline needs to be smarter. This is where we come in and this is how we are solving a critical piece of the problem.
Launchable journey started by wanting to help developers fail fast
Tests are the biggest bottleneck in software delivery cycle.
There is no way around running tests. You simply have to. They grow year over year. Your testing time continues to increase. A test failure will stop your bits from landing to production. A test failure found too late means days if not weeks in customers getting functionality to your customers. It’s almost like a tax - you do it because you have to.
What if there is a better way? There is and we found it - its called Predictive Test Selection.
We simply help find the right tests to run at the right time. This implies that developers can find issues much much earlier (as quickly as <5% of the entire test run). Risks are identified and fixed a lot earlier too!
The net-net is that software delivery speeds up.
Our ultimate vision is to take the data exhaust from every part of the DevOps infinity loop with an intent to help delivers find the right signals fast and deliver fast (more to come!). (read about why 645 Ventures invested in us in Series A here.)
Where does the story of Kohsuke & Harpreet go from here? A good ending here?
So the protagonists started a company aligned with their mission. They formed a crew with a bunch of like-minded people who are passionate about the mission and are busy executing. What’s the verdict huh?
“What’s your exit strategy - IPO or acquisition?” is a question that you often field as a CEO. Truth be told, a startup's journey never gets done & you never know; there is an element of gamble in it. That said as a CEO the focus is always on helping customers succeed. It is the journey and not the destination. That’s where we are today - finding like-minded people and helping them.
Have we crossed the chasm - not quite. Are we finding like-minded, forward-leaning folks? Folks, who want to build data-driven teams and deliver the highest quality software at the highest pace? Sure do.
All that said, I wouldn’t put it besides these two guys and their awesome team to do something meaningful for software teams everywhere.
(If you are one of the software development teams that could do with speeding up software delivery (by intelligently reducing tests) or a central devops team that cares about raising the state of art for your software teams, reach out to us on our discord channel, or our LinkedIn space or tweet at us - I am pretty sure we can help.)