Get lightning fast feedback on your commits

ML-assisted development. Less WIP. More in Production.


Used by Elite Engineering Teams


Find tests that are likely to fail, fast

Failing tests found by running 5-40% tests

A risk based approach powered by ML (and used by companies like Facebook) that prioritizes tests based on your commits.

We give you a legitimate reason to miss meetings and continue coding :-).



  • Not enough time coding
  • Issues found too late
  • Long testing cycles
  • Too much WIP. Not enough in production
  • Context switching on multiple streams of work
  • Old school dev practices in the team


  • Code more
  • Issues found early
  • Shorter testing loops
  • Less WIP. More in production
  • Less context switching. Focussed work
  • ML-assisted development

See how Manba sped up deploys to Heroku by 3x

Read how developers in Manba are happy that their deployments have sped up. See how we helped a leading car manufacturer in Europe speed up its dev team by 2x.


Why test everything?

Testing everything for every commit is a wasteful, old school practice


Complex codebases have long test cycles

You pushed your change in. Now you have to wait for your commit to run through a gauntlet of tests which takes forever.

80% of tests in the test cycle are pointless

You know 80% of tests are unnecessary for your commits but you just don’t know what 20% are the key ones. 

Pointless test runs implies delays to launch

Waiting hours and days as your code goes through the pipeline is just wasteful.


A data-driven approach to speed up dev cycles


A dynamic test subset based on commit metadata

We use commit meta-data to find the tests likely to fail and prioritize them to run earlier. Bottom line - if a test is likely to fail, we find them fast.

See how it works or head to our documentation.


Dev practices from elite teams

Up to 80% reduction in test cycles. No application rewrites

We have shrunk-wrapped Predictive Test Selection - an approach being used at companies like Facebook so that it can be used by any company.

Use the curves based on your data from your code, your tests to make data driven decisions to speed your development cycles.

the curve

Dead simple installation

A four line change to your build script

  1. Download CLI
  2. Instrument your build script. Read the getting started guide
  3. Let the model train for 3-4 weeks
  4. Speed up your test runs


Flexible to fit your delivery goals

We support multiple use cases: Shift-left or Shift-right across the testing pyramid.

We support greenfield, brownfield, monoliths and microservices - read more.

Read our FAQ to build a quick mental model of Launchable.


We support all developers

We integrate at the test runner layer - see the integrations and supported languages page for more information.


Secure by design

Code and tests remain on-premises. Only meta-data is sent to Launchable.

Data privacy & protection FAQ covers what we mean by meta-data (think git diff). You can run the audit command to inspect data sent to Launchable.

We can also engage your info sec teams.


We built Jenkins.

We are passionate about helping organizations deliver faster

co-CEO Koshuke Kawaguchi (ex-CTO, CloudBees - a DevOps leader) started the Jenkins project that is the bedrock of DevOps pipelines everywhere. Harpreet, our co-CEO was VP of Product for CloudBees where helped hundreds of teams through DevOps transformations. Our team has members from Atlassian Bitbucket, CloudBees, JFrog all leaders in developer productivity and DevOps.

Today, our passion is to help engineering build a data-driven culture to drive software releases faster.


Ready to engage?

Try it or talk to us if you want a guided POC


POC-ing Launchable

Evaluate if Launchable can help you

  1. Demo call Schedule a demo call to learn more.
  2. Discovery call Identify a project to see if we can make an impact.
  3. 4 Week POC You instrument (a couple of hours), sit back. Our ML learns and comes back with early results.
  4. Post POC decision call Go/no go. Typically a go go go! :-)


Our promise - respect your time

No pushy sales. Bring us if we show results.

  1. Identify a qualified project No POC if we don't think we can help
  2. Low people investment Your teams time is precious. POC typically takes hours to a couple of days to implement.
  3. Time bounded We show results in a 4 week POC
  4. Results on your data We show early results on your data to help you make data-driven decision to purchase us.

Your cart