How we signed our second customer

---

We started NOA in August 2023 with a love for automation and a plan to bootstrap. Our first customer was the ex-employer of Benedikt and Pius. And we’re genuinely thankful for that. It made the leap into self-employment a lot less uncertain.

But that’s why signing our second customer was what really tested us.

This was a company we had no prior connection with. A team of 50 employees producing custom products with large CNC machines. A classic German Mittelständler. As we quickly found out, their IT setup matched the stereotype. The ERP system was running on a server in the back office, completely isolated from the outside world. No APIs. No integrations. Just a lot of local windows and many, many manual steps.

We walked out of our first on-site meeting a bit bumped. We had expected challenges, but reality topped our expectations. Coming from the startup world, we were used to building automations around modern SaaS tools and cloud APIs. Here, that playbook was useless. Still, we managed to find a small, manageable use case: transferring salary data from the accounting tool into the ERP. It wasn’t the company’s biggest problem, but it was annoying enough to be worth solving. It took time, it was error-prone, and nobody liked doing it.

Our usual API-based stack didn’t stand a chance. The ERP was simply too closed-off. So we sat down and started digging. That’s when we stumbled across Microsoft Power Automate Desktop. It had the magic mix: capable of automating user interfaces and APIs, wearing the Microsoft badge, and already part of the client’s existing setup. For companies like this, it was the perfect tool. We felt like geniuses and were ready to plough through the entire German SME landscape like no one did before. We pitched a fixed-price offer: €2.000 for the whole thing split 50:50 between evaluation and implementation. A couple of days later, we met the client at a conference in Munich. Until then we hadn't from him regarding our offer. close to the end of the event, he pulled out the offer. Signed! There was not even our VAT-ID on the offer because we did not receive one yet. Anyway, our first sale and second customer. Maybe we were too cheap ;)

A wrinkled offer

We hadn’t worked with Power Automate before. But how hard could it be, right? As you probably expect, the actual implementation turned out to be anything but smooth. Power Automate Desktop is not exactly known for being intuitive or developer-friendly. We spent countless hours testing, debugging, and reworking flows. On top of that, we couldn’t develop directly in the client’s environment. So we improvised. We recreated their ERP in Notion. Yes, really. In Notion. We built a mini fake ERP to simulate the workflows and test our automations. Eventually, after many late nights, we had something that worked. We packed our gear and returned to the client’s office for the go-live.

Surprisingly, the final transition from our Notion mock-up to the real ERP went smoothly. Within two hours, we had the automation up and running. A process that used to take one hour now took one minute. It was a small change with a big impact. The client was happy. We were relieved.

That first project with them started small. But over time, it grew. Those 2.000€ turned into 100.000€. How that went and why we eventually banned Power Automate from our tool stack? Come back soon to find out.