Pour One Out For Heroku

You may have already seen this news about Heroku. While I would encourage you to read this on your own, the short of it is that Heroku is entering a “KTLO” or “keeping the light on” phase. Basically, no new innovations or features. They are just patching possible security issues. They aren’t even going to charge customers anymore (at least in the short term).

This tells me that Salesforce is effectively shutting down Heroku. Now I don’t know why this is the case but my suspicions are that Salesforce doesn’t want to be in the hosting business anymore.

We something similar back when Facebook bought Parse in 2013 only to shut it down in 2017. It makes sense as managing a hosting platform does take a lot of additional resources and doesn’t make sense if it isn’t your core business. That being said, Heroku is an OG.

What is/was Heroku

Heroku is one of the original Platform-as-a-Service (PaaS) providers. They were founded back in 2007 (Pre-dating Google App Engine by a year). To shed some context, in the early days of the cloud, everyone wanted virtual machines (VMs). This made sense as that was the hot tech in the 2000s and early 2010s.

That being said, developers didn’t care for it. They just wanted to host code. Heroku created an abstraction layer between VMs and the developer to simplify the deployment of applications to the cloud. I actually talk a little bit about this in my other blog.

Now creating a PaaS solution wasn’t their only contribution to technology. They also created the concept of the 12-factor app. Further, they open sourced the concept in 2024 during KubeCon North America. The also walked so that serverless can run in many ways.

I don’t think we would have serverless today if we didn’t have PaaS. It was a necessary stepping stone in redefining the cloud and making it more than just hosting VMs on someone else’s server.

Fun fact, in the early 2010s I was working on a startup (that got nowhere) that was modeled after Heroku (and dotCloud which eventually became Docker). I hated using VMs and bare metal servers (my first website was hosted on a server in my office area) and figured most developers did too.

Why the Change?

I have never been in a Salesforce boardroom nor do I know why they make decisions. But I can speculate. My theory is that they simply didn’t want to put money towards Heroku.

Unless you have been living under a rock, Agentic AI has been dominating the tech space for about a year and a half or so. There has actually been a lot of talk about Agents replacing SaaS (Software-as-a-Service). By the way, Salesforce has historically been a SaaS model.

Salesforce saw the writing on the wall so they went ahead and created their Agentforce platform. To put it simply, they created a way to host and scale agents. Salesforce has pivoted their business to lean heavily into this agentic AI space.

In this AI race, they likely didn’t really see a place for Heroku in their portfolio. Sure, they could repurpose it to be an agentic platform but then it would just compete with Agentforce. In my opinion, this is why they made the decision. They want to divert resources towards Agentforce and so something had to go.

Is This the End?

While I am sad to see this happen to Heroku, this probably isn’t the end of them. They may bounce back or be absorbed into Agentforce in some way.

Outside of Heroku, we have seen an explosion of Agentic AI platforms that leverage serverless architecture. I don’t think this is the end of PaaS or serverless but just marks an evolution or “passing of the torch”.

So let’s all pour one out for one of the OGs…

—Photo courtesy Aarón Lares on Pexels—