A Year of PAX: A Q&A with Founder Matthew Obey

It’s been about a year since PAX was introduced publicly. For this first-year reflection, founder Matthew Obey answered questions about what needed work, how AI use is managed in PAX development, and why human support still matters when a business depends on your software.

What did you expect a year ago, and what actually happened?

I expected the system to just work. It seemed clean and ready, but as we got closer to closing with our first customer, we found a number of things that still needed ironing out. It was all manageable, and we worked through them before bringing the customer on. We’d just felt ready before we actually were.

Once people started using PAX, their questions showed us where the system still needed to be more intuitive.

What has been the hardest part?

Getting seen. Manufacturing ERP is a huge market, and we’re trying to reach small manufacturers across the US.

I think these companies are underserved. They can end up buying an ERP with complexity they don’t need, then spend time dealing with headaches they could have avoided. The bigger providers have much more visibility, even when their systems may be more than a small operation needs.

We want to help those manufacturers streamline their operations and make their daily lives easier. Reaching them is still a challenge for us.

What moment made you feel PAX was genuinely useful to someone?

Users have shared their relief about how PAX performs compared with their previous systems. Hearing that has been one of the most encouraging parts of the year.

We originally built PAX because of our own frustrations with ERP software. I’ve written about that in the PAX origin story. Hearing similar relief from people using PAX gives me more confidence in what we’re building.

How has AI changed PAX development, and how do you manage its use?

We use AI to help plan changes, make them, and debug them. The individual driving a change is still responsible for the result. That’s always been the case.

AI is incredibly helpful for understanding large chunks of the codebase in little time. It can explain how different pieces of the system fit together much faster than reading through all the documentation yourself.

It’s also great for drafting repetitive SQL queries for reports. SQL is never fun to type out, and that assistance saves a lot of time. When we have a question about a resource or dependency, AI can often help us get an answer faster than working through a bunch of searches.

Asking it to build something outright is trickier. It likes to assume you’re Meta or Netflix and build things in the most complicated way. A basic database operation, such as adding or updating a record, can turn into what feels like 1,000 lines if you don’t watch it.

If software is getting easier to build, what do you think will make a small ERP company worth choosing and staying with?

I think the human relationship is the biggest part. Companies need to know they have someone they can reach out to about a system their whole business runs on.

If an error causes inaccuracies in the books and filings or reports are due yesterday, they don’t want an AI agent or an endless FAQ with no contact method listed. They need a human who can fully understand the problem and go in and clean things up.

What have you deliberately said no to this year: a feature, customization, expense or potential customer? Why?

Job sites. There are unique ERP requirements for contracting work, such as roofing or plumbing. It’s not something we’ve wanted to get into yet. We’re focused on manufacturers who build things in-house.

How do you balance getting another customer with improving the product for the people already using it?

We always prioritize current users over getting new ones. If someone finds a system bug, or requests something that would make PAX genuinely better, we prioritize that work first.

There’s enough time in the workday for both.

What are you proud of, and what do you want year two to look like?

I’m proud of how much more polished PAX has become, and of hearing users say they love the system and that it saves them time and effort. That was the goal: to make their daily work easier. Hearing that from the people using it means a lot to me.

For year two, I want PAX to have a broader reach and help more companies. We’ll also keep improving the system.

We’re adding mobile styles now, and most of the important modules are already in good shape on mobile. That work is still in progress.

I’d like to visit more customers and potential customers in person. I want to see their operations, how they do business, and how the people using the system get their work done, then bring what we learn back into PAX.