Why I'm Building My Own ERP Instead of Paying $35,000/Year for Software

For a long time, like a lot of operators, I accepted a basic software compromise.

You buy the platform. You adapt your team to it. You pay a pile of money every year. And even after that, it still does not quite fit the way your business actually works.

That is where I got stuck.

We were spending roughly $35,000 a year on software, subscriptions, workarounds, and tools that only partially matched the needs of the operation. Some parts were useful. A lot of parts were friction. Every gap created another spreadsheet, another manual step, another thing that lived in somebody’s head instead of in the system.

At some point, the question stopped being, “Which SaaS product is best?” and became, “Why are we still paying to bend our workflow around software that was not built for us?”

Generic software always has a ceiling

Off-the-shelf systems are often good enough to get started. They help early. They can save you from chaos when you need structure quickly.

But as the business grows, the mismatches become more expensive.

You start noticing that the platform is optimized for an average company, not your company. It handles the broad strokes, but the edge cases matter more and more. Purchasing logic, inventory movement, team communication, margin visibility, ecommerce sync, vendor nuances — the real work lives in those details.

If your system cannot match your actual operation, your people become the integration layer. That is inefficient, expensive, and fragile.

AI changed the equation

A few years ago, building custom internal software felt heavier. It meant larger budgets, slower timelines, and more dependence on traditional development cycles.

AI changed that.

Now the speed of prototyping, refining, and shipping internal tools is dramatically different. You can move from idea to working interface much faster. You can test workflows in real time. You can build around the actual business instead of settling for someone else’s default assumptions.

That does not mean AI magically writes a great ERP for you. It does mean the cost and speed curve are different enough that custom is finally realistic for more operators.

That is the opportunity I saw.

What I actually want from an ERP

I do not want a system that looks impressive in a demo. I want one that makes the business operate better on a Tuesday morning.

I want tighter purchasing visibility.

I want cleaner inventory control.

I want better margin clarity.

I want workflows that match how our team actually moves.

I want less duplication, less manual re-entry, and fewer moments where critical information depends on one person remembering something.

In other words, I want software that behaves like a real operating tool, not a billing relationship.

Why this matters beyond software spend

Saving $35,000 a year matters. Of course it does.

But the bigger prize is not the subscription savings. The bigger prize is control.

When you own the system, you own the roadmap. You are not waiting on a vendor to prioritize your problem. You are not stuck asking support whether your workflow can be hacked into somebody else’s product logic. You can improve the business at the system level as soon as you see a better way.

That kind of control compounds.

It improves decision-making.

It improves speed.

It improves accountability.

And over time, it can become a real competitive advantage because the internal operating engine is built around your exact model.

The real reason I’m doing it

I am building this ERP because I am tired of software that almost works.

Almost is expensive.

Almost creates hidden labor.

Almost forces talented people to spend time on nonsense.

Almost keeps businesses slower than they need to be.

If I am going to invest heavily in systems, I would rather invest in something that becomes more useful to us every month instead of more limiting.

Final thought

I do not think every company should build its own ERP.

But I do think more operators should question the default assumption that expensive SaaS is the only serious option.

If your business is mature enough to understand its own workflows, and if AI can help compress the cost and build time, then custom stops looking crazy.

It starts looking obvious.

I’m not building software for the sake of it.

I’m building an operating advantage.