Can you clone software with AI? According to a sharp new essay doing the rounds this week, increasingly yes, and that should worry anyone whose business is “we wrote the code, so we own the value”. We build Calendrz with AI agents ourselves, so we read it with more than passing interest. Our take: the code is getting cheap to copy. What the code has learnt is not.
The great software IP writedown
In The Great Software IP Writedown, Martin Alderson argues that software’s unusually fat margins never came from monopolies or network effects for most SaaS companies. They came from scarcity: too few people could build software to meet the demand. Agents now do much of that building, so the scarcity is going away. As he puts it, “The market always beats margins out of you when there is a shift in supply.”
He lines up the evidence:
- Cloning is trivial. Agents can rebuild an app from screenshots or by driving a browser. He points to clones of Excel and Photoshop, and says the Photoshop one arguably has more features than GIMP after 30 years of human effort.
- Switching costs are draining. Agents can export your data, reshape it and import it into a rival, even from apps that offer no export at all.
- The bottleneck moves to sales. Building is roughly ten times quicker; selling is as slow as ever.
- The writedown. If a buyer can rebuild a product in weeks, they pay for its customers and distribution, not its code. Without traction, IP is worth about what it costs to rebuild, which is “not very much anymore”.
It’s a writedown, not a write-off. But it’s a big one.
Why you can’t clone software with AI from a screenshot
Here is where our experience as builders comes in. Calendrz does one job: availability mirroring. When you’re busy in one of your connected accounts, it auto-blocks that time in all the others, without copying the private details, so you don’t get double-booked. You could describe the product on a napkin. You could probably clone its screens in a weekend.
But a screenshot shows the moment a meeting gets blocked. It doesn’t show what happens next week, when that meeting moves. And that is where most of the work has gone since our founder, Liviu Tudor, made the first commit in 2020. A few of the lessons from our own history:
- Recurring meetings once produced duplicate blocks. One weekly stand-up should not fill your other calendar with copies of itself.
- Rescheduled meetings are harder than new ones. Earlier this year we found a case where new blocks were created perfectly, but when the original meeting moved, its block stayed put. A check was comparing two different kinds of identifier, so they never matched. Everything looked fine on day one, which is exactly what a clone built from screenshots would copy.
- Cancelled meetings once looked like fresh changes on every run, so we kept “fixing” something that was already fixed.
- All-day events landed a day out until we taught Calendrz how each platform counts the end of a day, and that a full-day Google out-of-office is really an all-day event.
- Change notifications from Google Calendar and Microsoft 365 expire, arrive in bursts and sometimes arrive twice. We renew them before they lapse, hold them briefly to batch them, and make sure two servers never process the same account at the same time.
None of that is visible on a pricing page. All of it is the difference between a calendar you can trust and one that quietly lets a double booking through. Dolly the sheep was a perfect clone too, but she still had to learn her own way around the field.
So what is software worth now?
We think Alderson is right about the direction, and we’d go further. When a product is expensive and its features are easy to describe, people will simply build their own. A team running a handful of agents can now write the internal tool that used to justify a hefty subscription. High-priced SaaS that sells features rather than outcomes is the first in line to be replaced by something built in-house.
That is one reason Calendrz is priced the way it is: by tier, mainly by how many accounts you connect, never per seat, and low enough that a weekend spent rebuilding it would likely cost more than a year of using it. The value we sell isn’t the lines of code. It’s that six years of reschedules, recurrences and time zone surprises have already been met, fixed and tested, so you never have to meet them.
What to take from it
- If you build software, your moat is not the feature list. It’s the edge cases you’ve survived, the trust you’ve earned and the customers who stay.
- If you buy software, ask what you’re paying for. If the answer is “code anyone could regenerate”, expect the price to fall, or build it yourself.
- If you juggle several calendars, judge a tool by what happens when a meeting moves, repeats or gets cancelled, not by the demo.
Try it on your own calendars
Want to see how availability mirroring handles your messiest week? Connect your calendars to Calendrz and let it keep every one of them honest about when you’re actually free. Or read how it works first.
