Two years ago I would have called this a mistake.

When we started working on our internal workflow management tool Promotheus (Promo + Prometheus), I had flashbacks to my first agency job where they built what was clearly a prototype of a workflow management tool and the execs said, “ship it”. The pitch was great but I thought we risked either tethering some poor soul to it permanently, or shipping something that did a worse job than a dedicated SaaS product.

Two years on, and it is the beating heart of the company that touches every part of the business. The compounding effect of working on something like this in-house is that it’s tailored specifically to our needs, and it’s almost impossible to replicate as it has grown out of every decision and pivot. The addition of agentic coding has also brought big changes:

I’m not one of those AI optimists who claim there’s no longer a case for SaaS companies, but the use and selection thereof has become much more intentional. This year alone, we’ve in-housed our CMS, Jira, and Digital Asset Manager. In some cases, we weren’t using the full set of features, so it’s not as though we have to duplicate the entire offering; just the slice that matters to us. The benefits don’t just stop at money saving, or first-party integration. They also mean fewer places to manage user roles, better security posture, and a product that changes based on how you work, rather than changing the way you work.

Build vs Buy is still a relevant calculation, and we use SaaS strategically in places like error monitoring (the world doesn’t need another OpenTelemetry wrapper) and accounting (integrates with the ATO. Enough said), but it is notable that things that were an obvious “Buy” 18 months ago have become an obvious “Build”.

What are you in-housing today that you wouldn’t have touched in 2024?

Cross-posted from LinkedIn.