How Heroku Is Positioned To Help Ops Engineers in the GenAI Era
The new generative AI-based coding assistants are truly remarkable, according to Bob Wise, CEO of Heroku. But at this point, “The effect of having all of that amazing code generation done is a little bit like adding an infinite number of interns to your development team,” he said in this episode of The New Stack Makers.
The GenAI revolution is, thus far, not making life any easier for engineers responsible for testing, deployments and getting software into production, Wise told Alex Williams, founder and publisher of TNS, and host of this episode of Makers.
The rise of GenAI coding tools, he said, reminds him of the impact Kubernetes had — with the difference being that Kubernetes was aimed at automating the backend rather than the frontend of software development.
“One of the reasons why Kubernetes was successful is that developer experience was not really the thing,” Wise said. “It was really about operator experience, because most teams, most companies, spend 90% of their effort and time on Day 2 things — on patching and maintenance and troubleshooting and all of the things, as opposed to development piece.”
Kubernetes “really attacked the operations pain fundamentally, which is one reason why people saw so many advantages,” he said.
By contrast, code generation tools may help developers write code faster, but so far, they’re not helping operations to move faster. And “most teams, most organizations, are not really yet practicing full continuous deployment.”
A Winning Bet on Postgres
Wise pointed out a number of decisions made by Heroku, a cloud Platform as a Service (PaaS) that have positioned it to take advantage of the ways in which GenAI is changing IT needs.
“We sort of help with the eat your vegetables, take your vitamins, do your exercise part of the software life cycle, so that you can actually take advantage of all of these new productivity tools that are in the development space,” Wise told the Makers audience.
“But it’s a tough chasm to cross to get from being more human-based to being more fully automated.”
He added, “Opinionated systems like Heroku are going to see a lot of growth here over the next five years, as all of these generative tools get better and better and better.”
Heroku, acquired by Salesforce in 2010 (which controversially shuttered its free tier in 2022), began as a container orchestrator and has since introduced database services, creating CI/CD features and placing a bet on Postgres and Kafka.
With Postgres seeing a surge in popularity due to its vector database support, a benefit for teams running GenAI workloads, Heroku’s bet on Postgres has paid off, Wise said.
“Postgres has really emerged as core, standardized database technology,” he said.
He added, “The Postgres community has been pretty amazing at responding and building extensions and other things on Postgres that have kept it in the forefront of being the standard.”
Up Next: A Kubernetes Rebase
Heroku was created before the dawn of Kubernetes. But now that K8s has matured, Wise said, “We decided that the best thing for us strategically was to use Kubernetes underneath the covers, essentially to replace a lot of the code that we have been using to do the same thing. That’s a big migration effort for us. There’s a lot of expertise that needs to be built.”
“We’re a long way down that road now. That stuff will be hitting production later this year … it’s definitely been an interesting voyage. But we felt like it was a strategic necessity for us to really converge on the industry standard.”
He added, “You can sort of think of our new Kubernetes rebase as being the Heroku opinion as to how you should run on Kubernetes.”
Check out the full episode for a deep dive into how Heroku has evolved and what’s on the horizon.
