#10 The Week I Held My Breath: Moving to a Real Server

I had decided to move my tool to a real server. The only problem was I had no idea where to begin.

My tools all ran on spreadsheets back then. They worked, but one of them — a search tool I built to find supplies in our hospital storeroom — had gotten painfully slow. So I asked Gemini how to put it on our office NAS instead. (A NAS is just a storage box that sits in the building and serves files to everyone on the network.)

The answer was a long list. Settings to configure, things to move, permissions to set. And to do any of it, I needed the admin account.

I had never even seen that account. I worked in a therapy room. Admin access to the machine the whole hospital relied on was about as far from my world as it could get.

There was only one way forward. Go ask for it.

Asking for the keys

I went to the manager in the planning office.

I told him my tool was too slow, that I wanted to try moving it to the NAS, and that I'd need the admin login and password to do it. Looking back, it was a bold thing to ask. A therapy-room employee, requesting the master key to equipment the entire hospital used.

He hesitated. He had good reason to.

Years earlier, that same NAS had been attacked. Someone broke in from outside, locked the whole thing, and demanded money or the files would be gone for good. In the end, the hospital paid to get its data back. After living through that, of course he was nervous about someone wanting to put more stuff on it. What if it gets attacked again?

So I told him this: I'd build it to run only inside the hospital, only for staff. If we never opened a door to the outside world, there'd be nothing for an attacker to come through.

I'll be honest — that was a reckless thing to promise. Back then I barely understood security at all. But it happened to be true in that moment, and he handed over the account.

A week of bracing for the worst

What followed was a week of pure tension.

I'd been told the old attack probably got in because the admin account had been left logged in. So every time I touched that account, my nerves were on edge. What if I press the wrong thing and erase the records of the entire hospital? When I finished for the day, I checked two or three times that I'd actually logged out.

It felt like a stranger fumbling around inside someone else's home — and not just any home, one that had already been burgled once.

The work crawled. I followed Gemini one line at a time. Configure this. Move that. When it broke, ask again. Solve one thing and the next wall appeared. I clung to it like that for a week.

The moment it worked

A week later, the moment came to open the tool for the first time on the new server.

I took a breath and ran it.

The screen came up. Not just came up — it snapped open, instantly. I typed a search. The tool that used to make me wait and wait spat out the result the second I pressed enter. Out of hundreds of thousands of supply items, the one I wanted appeared right away, like a trick.

It was absurdly fast.

I sat there staring at the screen for a while. A week of clenched nerves let go all at once. It worked. It actually worked.

The math that broke something open

Something opened up in me that day.

There was one more worry I'd carried in. Our NAS held five terabytes total, and every employee used it for photos and documents. I figured my tool would eat up space and crowd everyone else.

It turned out to be nothing. Data made of text barely takes up room. All those hundreds of thousands of supply records, added together, came to less than one megabyte. Lighter than a single photo.

Faster on the server. Safe because it stayed inside the building. And it took almost no space. Once I had confirmed those three things with my own eyes, the next thought followed on its own.

Then I could just move everything I've built over here.

If there's one thing I'd hand to you from that week, it's this: proving something works once, in the smallest possible version, is what gives you permission to think bigger. I didn't need a plan for all twenty tools. I needed one tool, on one server, working.

Every program I'd been running on spreadsheets suddenly had a way up to the next level. That one small win — moving a single tool — turned out to be the starting line for everything that came after.

Of course, back then I didn't yet know that moving the rest wouldn't be nearly as easy as I imagined. But that's the next story.

This is part of From Zero to Ship — the honest logbook of a non-developer who shipped 20+ working tools with AI. If this resonated, the rest of the series is there.