Humn.ai uses machine learning-based models to quantify commercial fleet exposure at individual insured asset risk in real-time. RiskOS processes diverse streams of data that power dynamically-updating risk assessment algorithms, pricing, and product offerings, tailored to the actual usage of individual insured assets.
Scaling Production Deployments
Alberto Romero, CTO at humn.ai, wanted to give developers access to on-demand deployment pipelines. He originally built a software delivery process using Jenkins and Ansible, but these tools wouldn’t scale for modern applications in the cloud. Humn.ai was beginning to migrate it’s legacy monolith EC2 machines to Kubernetes microservices, and the Jenkins/Ansible pipelines made it hard to deploy, debug, and verify configurations in production.
Alberto needed to find a modern deployment solution. His search led him to evaluate Spinnaker and Harness.
The Trials of Spinnaker
Alberto initially conducted a two-month Spinnaker POC.
The installation and administration of Spinnaker required advanced Kubernetes knowledge not possessed by humn.ai’s developers, and without a dedicated DevOps team, there weren’t enough engineering resources to decipher Spinnaker’s complexity.
After two-months, the team was unable to implement Spinnaker. Alberto began searching for simpler deployment tools, which lead him to try Harness.
Production Deployments In Days Using Harness
Harness’s intuitive setup allowed Alberto to set up a pipeline and make his first production deployment in under a week. Alberto asked two of his engineers to familiarize themselves with Harness, who later completed their first deployment within a week.
Harness’s ease of use and customer success on-boarding allowed humn.ai to deploy 117 services in just two months. Deployment velocity has increased without manual intervention because developers can now deploy on their own using the Harness self-service platform.
Alberto’s team no longer worries about maintaining deployment pipelines and instead is able to focus on development.