Jenkins creator takes on the next frontier in DevOps -$9.5M
Where It All Started and the New Heights Launchable Has Reached Today
Key Takeaways
The problem: Somewhere in mid-2019, Harpreet and Kohsuke (the creator of Jenkins) were talking about the numerous struggles that organizations were facing delivering software at high quality.
The solution: Build Launchable to compare code and past test history to build a machine learning model.
Today, Launchable announced that they have raised $9.5M led by 645Ventures.
Hi
I am Harpreet, co-CEO at Launchable and thought that since everybody loves an inception story, I will start there.
Harpreet, Kohsuke - the day Launchable was born
The problem: Devs are not getting the right signals at the right time
Somewhere in mid-2019, me and Kohsuke (the creator of Jenkins) were talking about the numerous struggles that organizations face delivering software at high quality. Every software team that we talked to is challenged finding the right balance between speed and quality. Developers are frustrated because they are waiting hours for feedback. The key problem is that they aren’t getting the right signals at the right time which slows them down or leads to low quality code in production. Bad code pushed fast is a recipe for losing customers. This sentiment is echoed by multiple stakeholders.
Churn is the affliction of the SaaS business -- a leading VC
Walking a tight rope between speed and quality (credit: Unsplash)
DevOps transformations take too long to bring change
In the last 10 years or so we have helped countless organizations go through DevOps transformations but felt that these transformations were too slow. Waiting months before seeing any tangible reduction in cycle times is painful.
With the adoption of DevOps practices, there is a tsunami of data coming out of the developer pipeline and engineering teams are not fully utilizing this data. Kohsuke’s vision is to up the state of the art in DevOps to drive a data-driven DevOps practices in organizations.
Key Insight: Tests are a bottleneck and nobody is working on optimizing them
Kohsuke’s insight was that the biggest bottleneck in the development process was testing/quality. The bottleneck gets exacerbated with Automation. The infinity loop is considerably lop-sided on the testing side. Teams have no option but to go through the entire gauntlet of tests & test suites slowing cycle times down.
As automation grows, devices, platforms grow the tsunami of data continues making the infinity loop even more lop-sided.
The Approach: Using data for “Smarter Testing for Faster DevOps”
Our solution was to take advantage of the data to drive intelligence in the testing layer. We wanted to bring intelligence to testing.
Why do I have to run all the tests? – every developer
We wanted to emphatically answer this question with a “You don’t have to run all the tests” - an assertion backed by data driven insights and action.
The Solution: Predictive Test Selection
Our solution is fairly simple to pitch. We look at your code and your past test history to build a machine learning model. The next time changes come in, we can predict what tests to be run to find issues faster. By finding issues faster, we can help organizations drastically reducing cycle time without sacrificing quality.
September 2019 Launchable is born
We talked to our friend Dharmesh Thakker from Battery Ventures and our hypothesis and problem area really meshed with what Battery was hearing from the market. Dharmesh thought that this space is ripe for innovation and our approach really excited him. He encouraged us to start a company. John Vrionis at Unusual Ventures quickly came on board.
We raised a Seed round and brought in some power hitters in the industry like Sri (CTO of Atlassian), Sacha Labourey (ex-CEO of CloudBees) who resonated with the problem and vision. See our investors here.
The early team at Launchable
Along the way, we had valuable input from our advisors Max Schireson (ex-CEO MongoDB, Battery), Chiraag Deora from Battery Ventures.
Where we are today: A $9.5M raise led by 645 Ventures
We thought about our next phase of growth to help our customers and deliver on our vision. We wanted investors who truly understood the pains in development pain, the challenges raised by automation and DevOps.
I met Alessio Fanelli last year who reached out to me because he was bummed about slow tests in engineering teams. He emphasized developer delight as one of the key things he looks into while investing. I was blown away when he signed up for the product with a Github repository, tried the product and started a bunch of feedback our way. Nnamdi Okike was a kindred spirit. We all instantly connected.
Today, I am very excited to announce that we have raised $9.5M led by 645Ventures. Our friends at Battery and Unusual have upped their investment in the Series A. GoingVC have come in because they love our vision as well
Our Product - Predictive Test Selection + Flaky Tests Insights “apps”
In the last year and a half, we have been furiously building the platform.
Today, we have delivered two applications the platform.
Predictive Test Selection (spoken earlier) which shows a reduction of 40-80% reduction in cycle times in weeks. We have had some compelling big name industry stalwarts using the platform and our solution in production (see case studies).
Training the Launchable Brain
Flaky Tests Insights which tackles the problem of flagging unreliable tests.
Flaky Tests Insights
Extensible platform: Designed for a heterogenous enterprise
We learned quickly that teams have heterogenous environments - languages, framework, application types. Thus, we have built an extensible platform that can consume data from multiple languages and test platforms. We can easily add new test platforms (in a couple of days each). You can read more on the various integrations on our docs site.
Support for multiple languages and frameworks
Emphasis on turn-key adoption: Days and not months for transformations
We have spent a considerable effort to make our solution turn-key. You can effectively connect to us in about 4 lines of code changes to the build script. You can start using the model from day 1 or wait up for a few weeks to work the fully trained model.
The real game changer here is that as team adopting Launchable - you connect to us and can go away and do your work and soon enough you start getting feedback faster and reduced cycle times. You are not tied to a long transformation project.
Where are we heading next and What it means for our customers?
Our Vision: Continuous Quality for developers
We will like to up the state of the art for developers to bridge the gap between speed and quality. We want to examine the data that is coming out from code changes, tests suites and not only provide insights and actions that help them deliver software fast and high quality. We will do so by delivering “apps” based on our intelligence platform (today: we have two & more to come).
Better yet, let’s use an analogy (as Kohsuke really likes to lead with them).
Today, software delivery funnel & testing is like a stick shift car. You push code and then run through a gauntlet of test suites (effectively moving into higher gears as you move right). However, there might be tests right at the end of the funnel that are suited to run earlier. Launchable wants to examine that entire funnel make the appropriate decision on what tests need to be run where and automatically take that decision. Essentially moving from a stick shift to an automatic car.
Ultimately, we want fix the lop sided DevOps infinity loop and make quality a part of the conversation where it isn’t a bottleneck for developers but an enabler. We want to bring “Continuous Quality” up front and center in a developers life.
Implications for our customers
Our focus in the near future will continue to making our platform better, reliable (we handle a good magnitude of tests already) and easier to use. We will continue building new applications on the platform for dev teams. Dashboards to help engineering leaders to clearly show the impact that they have on the companies customers and consequently the top and bottom line. (As an aside: I to thank our early customers, design partners for engaging with us (you know who you are) - Thank you!)
Implications for developers
We think every developer out there can benefit from getting some time back in their lives. Today, a developer can sign up on the web app and try the product and start using it. To make it easier for developers - we have a free tier (for teams under or equal to 5) and $20/developer/month for small teams under 40 with a 60 day trial (see pricing). We will continue to invest in more platforms and making it easier for developers to use the product.
Today, we are also announcing that the product will be free for qualifying open source project (talk to us) to enable it for your open source project.
Hiring
We are a values driven organization and have strived to build an organization that lives and breathes these values. In a way, Launchable was started because of our “Be the change you seek” value. We are looking to grow the engineering, marketing, sales and customer success teams.
Help customers get successful: We have staffed up our customer success team to enable our customers and prospects to get successful faster. After all, customer success is in line with one of our core values “make customers our raving fans”. We will continue to make progress on this front.
Educate the market: Development teams don’t really look into testing as they try to improve cycle team because they have conceded the space “as not an easy problem to solve”. We want to bring “Continuous Quality” into the developer vernacular and make them aware that problem is a “low hanging fruit” with Launchable’s approach. To that end, we are building a marketing team, hiring a marketing leader, growth marketer and evangelist on our team.
Bring more customers We are building out our sales team to talk our prospects and reach out to people who have cycle time problems due to tests.
Build more product: Drive towards the vision to build a Control Pane for Quality. Thus, we need stellar engineers, data scientists and data engineers.
Asking for help
It takes a whole village to grow a child.
To that end, it takes a whole lot of friends and well wishers that can help a startup grow into its teens. If you can help - referring people who have the problem, pointing to stellar players we can hire - we would love to hear from you.
Meanwhile, jump over to our discord channel and join us.
– Regards
Harpreet (& Kohsuke) - founders, Launchable Inc