I rebuilt Notion in 5 hours

I used to think the SaaSpocolypse / SaaSmageddon / the death of SaaS was overblown.. until last Saturday.

Who has the time to rebuild something like Notion?

Apparently, me. And it was far easier than I expected.

More importantly, it is only getting easier. What took 5 hours soon becomes 5 minutes.

Over the past few days, I have seen the same idea gain momentum on X. Guillaume at Vercel, David Senra on TBPN and even David Friedberg on All-In have all spoken about this. Friedberg even said he had rebuilt an internal SaaS app at 11:30pm on a Sunday night.

SaaS won the past decade

Remember 12 months ago when building software was slow, expensive and hard?

In that era, paying a monthly subscription was obviously more economical than hiring people to build custom software yourself.

That was the foundation of SaaS.

But AI coding agents are collapsing the cost and time required to build software.

Cloning Notion in 5 hours

This idea has been building in my head over the past few weeks due to:

• Opus 4.6
• Flare's team growing quickly
• Our Notion bill was becoming noticeable. We were paying more than $500 per month.

With Cursor, I supplied an incredibly simple initial prompt and had something usable within 15 minutes.

From there the hard part was not coding, it was deciding what mattered.

Features like search, deletion, guest sharing and sub page behavior required more detailed prompting to get production-ready.

It actually turned out to be a fun Saturday because of all the custom Notion features that were suddenly unlocked.

For example, I added teams to Motion and Flare-specific logic for assigning users. I added stronger deletion and locked down critical documents that should never be removed. I added AI drafting tools so going from 0 to 1 on a new file, using our internal document structure, became smoother.

I packaged the project up as motion-oss, a public open-source repo for anyone - or any agent - to use.

On Sunday I spent another 5 hours polishing and by Tuesday our team of 30 had been migrated onto Motion.

Every founder I have told this story to has had the same response: "What the fuck."

Who dies first?

The most exposed tools in my opinion are the ones whose value lies within the UI, usability and design layer.

Think: Notion, Linear, Asana, Airtable.

That does not mean they disappear overnight. But it does mean their value prop starts to look very different when a motivated founder can quickly produce a clone.

It is worth adding: these companies have incredible founders and I bet that many are extremely capable of building new features to stay relevant. But change is required.

Another argument I have seen is that seat-based models will suffer. I think that is directionally true, but I am less convinced it is the whole story. If we end up with thousands of agents using software on our behalf, then usage may explode rather than shrink. Yes, pricing models will need to change, but I do not think seat-based will have to go away.

It is not a full blown extinction

There is going to be some serious SaaS winners here.

Particularly those that act as a system of record (QuickBooks), have compliance depth (Deel), proprietary data (Snowflake), network effects (GitHub), unique distribution (Shopify) or quickly become agent-first.

These tools will likely see a surge in usage as agents begin to increasingly automate human jobs, operating on top of the existing human-used systems that are too cumbersome for the agent to rebuild (at the moment).

One open question I think about is whether this class of SaaS apps will embrace their new agent customers. If they do not, they run the risk of losing ground to competitors which do.

The real takeaway

The SaaSpocolypse is not murdering all SaaS.

It is killing apps that were renting out coding scarcity.

What we are really watching is a repricing of what code is worth.