The insight about horizontal vs. vertical SaaS is spot on,,, it turns out that charging per seat to manage administrative text is highly vulnerable to AI, while proprietary workflow data remains incredibly secure.
This is the cleanest version of the seat-math argument I've read, and the durable-middle point is right. The one place I'd push: I think the horizontal/vertical line is a proxy for the thing that's actually doing the work, and occasionally it mislabels. The real question isn't whether software is embedded in a regulated or physical workflow — it's whether an agent can route around the function entirely.
Most horizontal seat-tools are exposed because the agent absorbs the workflow. But some vertical tools are durable for a softer reason than they look: the moat is real, but the revenue mechanism sitting on top of it can still get severed independently.
Salesforce is the tell. You file it as exposed at 3.5x, and directionally that's right — but its problem isn't that it's horizontal. It's that the agent it's selling does the work the seat was licensed for, which breaks the join between a genuinely durable switching cost and the per-seat line that converts it to revenue. The moat doesn't move; the meter does.
So the audit question I'd add to your three: not just "could my team build this," but "does the value here depend on a human being the one who acts?" That second test catches some vertical tools your multiple-based version would wave through — and spares a few horizontal ones it would condemn.
I've seen this exact shift happening with a lot of teams lately.
The moment you realize a custom internal workflow can replace dozens of expensive user seats for a fraction of the API cost, the math on traditional software completely changes.
“Vertical SaaS platforms deeply embedded in regulated industries or physical workflows: insurance platforms like Guidewire, field service systems like ServiceTitan, fleet and IoT platforms like Samsara, vertical commerce platforms like Shopify.”
totally agree. these are incredibly safe bets. some of these might be replicable with a vibe coding session on the front end, but the scale and real world infra is not.
Per-seat pricing for horizontal tools was always a bet on friction lasting forever.
It really was! And it paid off for a while, but not forever
So many amazing opportunities right now.
The insight about horizontal vs. vertical SaaS is spot on,,, it turns out that charging per seat to manage administrative text is highly vulnerable to AI, while proprietary workflow data remains incredibly secure.
exactly! That’s futureproofing your business
Hubspot at 8B valuation with estimated 3.6B in revenue in 2026..
That’s a good one! And one I personally use!
the build-it-yourself part is more within reach than people think
im building my whole platform solo right now with ai, no dev team
the stuff that used to need an engineering org just doesnt anymore
Cory, great point! It really is within reach! It’s the kind of thing you need a taste of, and you see what a possible
This is the cleanest version of the seat-math argument I've read, and the durable-middle point is right. The one place I'd push: I think the horizontal/vertical line is a proxy for the thing that's actually doing the work, and occasionally it mislabels. The real question isn't whether software is embedded in a regulated or physical workflow — it's whether an agent can route around the function entirely.
Most horizontal seat-tools are exposed because the agent absorbs the workflow. But some vertical tools are durable for a softer reason than they look: the moat is real, but the revenue mechanism sitting on top of it can still get severed independently.
Salesforce is the tell. You file it as exposed at 3.5x, and directionally that's right — but its problem isn't that it's horizontal. It's that the agent it's selling does the work the seat was licensed for, which breaks the join between a genuinely durable switching cost and the per-seat line that converts it to revenue. The moat doesn't move; the meter does.
So the audit question I'd add to your three: not just "could my team build this," but "does the value here depend on a human being the one who acts?" That second test catches some vertical tools your multiple-based version would wave through — and spares a few horizontal ones it would condemn.
Man thanks for the thorough comment. Salesforce is a great example here, great addition!
I read this like a tool I can use for mine AI sessions. Mucho thanks.
For sure!
I've seen this exact shift happening with a lot of teams lately.
The moment you realize a custom internal workflow can replace dozens of expensive user seats for a fraction of the API cost, the math on traditional software completely changes.
it really does! It’ll be interesting when that thinking becomes common
The interesting question now is how much software you’re paying for out of habit.
Some of those tools are basically expensive furniture. Nobody likes them and somehow they’re still in the building.
Haha yes! I do think slowly, companies will begin to question each one (or at least I hope they do)
“Vertical SaaS platforms deeply embedded in regulated industries or physical workflows: insurance platforms like Guidewire, field service systems like ServiceTitan, fleet and IoT platforms like Samsara, vertical commerce platforms like Shopify.”
totally agree. these are incredibly safe bets. some of these might be replicable with a vibe coding session on the front end, but the scale and real world infra is not.
great read!
Nice!
Thanks!