The phone call
It was March 2019. A photographer I’d worked with on and off for years called on a Tuesday afternoon. Her voice was level in the way people’s voices get when they’ve already been crying for an hour and now they just need someone to help them think.
The drive on her desk had started clicking Sunday night. She’d turned it off, turned it back on, heard the clicks, turned it off again. By Monday morning it wouldn’t mount. By Tuesday afternoon she was calling me.
Nine months of shoots. Weddings. A product campaign she’d been chasing for two years. Raw files, selects, delivered galleries, the in-between scratch work that clients never see but which is how the good ones get made. All of it on one drive, because the second drive she’d been “meaning to set up” had been sitting in its shrink wrap on the shelf behind her desk the entire time.
We tried the usual things. A recovery lab quoted her $2,800 for a best-effort attempt and no guarantees. She paid it. They got back maybe sixty percent, most of it corrupted. Clients were understanding in the way clients are sometimes understanding, which is to say some of them were and some of them weren’t, and the ones who weren’t made the next six months of her life worse than they needed to be.
I thought about that call a lot over the next few years. Not because it was unusual. Because it wasn’t.
The tools we had
In the months after, I went looking for what I would recommend the next person who called. I’d been helping friends with computers since I was twelve. I’d run small IT for creative shops. I thought I knew the landscape. Here’s what I found.
Time Machine is fine for what it is. Apple ships it, it works, it’s free, and if your worst-case scenario is a single laptop and a single external drive that never leaves your desk, it’s a reasonable answer. It is not an answer for someone with an active client workflow across three Macs and a pile of shoots that live on rotated drives. It has no cloud tier. You can’t see across machines. If the drive it writes to sits next to the Mac it backs up, any disaster that takes the Mac takes the backup.
Backblaze Personal was, for a while, the thing I’d been recommending to friends. Unlimited cloud backup for one Mac for a reasonable monthly price. It is a real product that saves real people from real disasters. But it is built for one computer at a time, rotated external drives confuse it, its treatment of user-held encryption keys is thinner than a working professional should accept, and the restore experience when you actually need it has not kept pace with the rest of the product. I’ve watched that restore experience in anger. It’s fine. It is not what the photographer on the other end of that call needed.
Enterprise backup tools — Druva, Rubrik, the big names — are priced like it’s 2028 and built like it’s 1998. They are solving a different problem for a different customer. Asking one of them to handle a four-person photo studio is like hiring a fleet management company to change the oil on a single van.
Arq was the closest thing to what I wanted to recommend, and I did recommend it for a while. It’s serious software built by serious people. It respects user-held keys, it supports multiple destinations, it understands that real backup isn’t one-size-fits-all. But it isn’t Mac-native in the way Mail or Photos or Finder are Mac-native. The configuration surface area asks things of a user that most creative professionals should never have to think about. Every time I set it up for someone I’d brace for the call two weeks later about a setting they’d clicked.
Carbon Copy Cloner is a great tool. It is a clone tool. It is not a backup product. Cloning is a thing you do once, or on a schedule; backup is a posture you hold continuously. Conflating the two is how people end up with a pristine clone of their drive from six weeks ago and nothing in between.
Each of these products was built for an audience. None of them was built for the audience I was trying to help.
What the product had to be
Sometime in 2023 I wrote down seven things on a notecard. If a product did all seven, I’d recommend it and stop trying to build one. Nothing on the market did all seven.
- Mac-native. Menubar app, native install, real macOS conventions, not an Electron wrapper around a Windows product.
- End-to-end encrypted with user-held keys. Your passphrase, your data, no backdoor, and the service operator cannot read your files.
- Continuous. Not on a 24-hour cron. Not “when you remember.” Continuously, in the background, while you work.
- Versioned. Every backup is a point in time you can restore from, not a single sync state that mirrors deletion.
- Multi-destination. Your cloud. Your external drive. Your bring-your-own-storage bucket. At the same time, from one configuration.
- One account for multiple Macs. A studio laptop, a desktop, a partner’s machine. One login, one view of the fleet.
- Testable restore. You should be able to practice the restore before you need it, and the product should encourage you to.
Seven is not a lot. None of them is exotic. But combined, they describe a product that didn’t exist.
Why now
I don’t believe “why now” is a marketing question. It’s a real one. The old answers stopped scaling for concrete reasons.
Consumer MacBook Pros shipped past 2 TB of internal SSD as a standard option. A working photographer on a modern body is moving hundreds of gigabytes a week in raw files. Video professionals moved from 1080p to 4K to 6K and now 8K on consumer sensors. The working set of a creative person is not what it was when Time Machine shipped in 2007.
Cloud storage finally got cheap enough to take seriously as a backup destination, not just a sync destination. The price per terabyte per month at reputable object-storage providers is in the five to six dollar range in 2026 — cheap enough that running a continuous, versioned, off-site copy of your working set is no longer a luxury tier. It’s the table stakes.
Ransomware in 2026 is not a Windows problem. Platform-crossing variants enumerate mounted external drives, network shares, and in some cases synced cloud folders. A backup strategy that assumes the attacker can’t reach your backup drive is a backup strategy that was written in a different decade.
And creative work shifted. A working photographer in 2026 has more machines, more shoots per year, more file sizes per shoot, more clients expecting faster turnarounds, and less tolerance for a day lost to recovery. The cost of a disaster went up. The tools didn’t keep up.
What we chose
We made some architectural decisions early, and the ones we think matter most we’ll say out loud.
We chose an open-source backup engine with a public command-line tool as our foundation. If macup the company stops existing tomorrow, your backups are still readable by anyone with the passphrase and the engine. You are not trapped inside our format.
We chose to offer a managed cloud tier with no egress fees. When you need your files back, you need them back. Paying a per-gigabyte ransom to retrieve what’s already yours is a pricing model we refused.
We chose to offer bring-your-own-storage as a first-class citizen, not a hidden power-user option. Your studio may already have an object-storage contract. Your IT person may have opinions. Good. Point us at it, hold your own keys, and we’ll do the hard work around it.
We chose a menubar app that tells you what it’s doing at every moment, in plain English, because a backup product that doesn’t is a backup product that teaches its users to ignore it.
And we chose to test the restore flow in front of real users six times before we shipped it. Because the backup is the easy part. The restore is the part that actually matters, and it’s the part every other product in this space has underinvested in.
The call I wanted to give her
I can’t give that photographer her nine months back. No one can. The drive was the only copy, and the only copy is no copy.
What I can do is build the tool I wish I’d been able to hand her in 2018, before the drive clicked. Mac-native. Continuous. Encrypted end to end with her keys. Running to her cloud and her external at the same time. Restorable, and rehearsed, before she ever needed it.
That’s macup. If you know someone who is one drive failure away from the call I got, try it free for fourteen days. Then go make them install it.
We can’t reach everyone in time. But we can reach the next one.