Every year for SaaStr AI Annual, we have to distribute 4,000+ parking passes.
That alone is a pain. But it’s worse than it sounds:
- There are 3 different parking pass variants (different lots, different access levels)
- Each pass has to be tied to a specific attendee by name
- That’s 12,000+ possible permutations across the attendee list
- And all of it has to happen in the last 10 days before the event
For 13+ years we’ve dealt with this the same way. A mess of spreadsheets, copy-paste, mail merges, and eventually some poor human on the events team sending thousands of emails late into the night. Every year. Every event. Every time.
This year, Amelia (our Chief AI Officer) vibe-coded a mini-app in Replit with Claude Desktop.
60 minutes. 4,000 personalized emails. 12,000 permutations. Done.
Every attendee gets a custom email with their correct pass attached, their correct lot listed, and the right days covered. Clean, professional, zero humans involved beyond the hour it took to build the thing.
Here’s what’s actually disruptive about this
- You could never pay a developer to build this. Think about the economics. An outside dev shop would charge $15K–$30K to build something this specific, and by the time they understood the 3 pass variants and the attendee data model and our weird edge cases, you’d already be past the event. Internal dev? Same problem plus opportunity cost.
- And no vendor would build it either. The TAM for “parking pass distribution software for annual B2B conferences with 3 lot variants” is roughly $0. It does not exist as a SaaS category. It never will.
- But we still need it. Every year. And every events team everywhere has some version of this problem. Thousands of tiny, specific, idiosyncratic workflows that no SaaS company will ever build for you, because the market for each one is you and maybe 4 other companies.
Now we just build it ourselves. In an hour.
This is what “N=1 apps” actually means
An N=1 app is software where the total addressable market is exactly one customer: you. Your one workflow, your one data shape, your one weird set of constraints.
For the last 30 years, N=1 apps were economically impossible. You’d need a developer, a stack, a deploy pipeline, a maintenance plan. None of that math worked for a tool that one person uses 4 times a year.
That math just flipped.
When a non-engineer (or in Amelia’s case, someone who stopped writing code professionally years ago) can ship a working internal tool in under an hour with Claude + Replit, the floor on what’s worth automating drops to near zero.
What this means for SaaS
A few things worth sitting with:
1. Every annoying workflow is now a candidate for automation. Not the big ones — those already have vendors. The small, annoying, repetitive, once-a-quarter ones that were never worth a tool before. Parking passes. Vendor badge assignments. Speaker green room confirmations. Sponsor booth directions. The long tail of event ops alone is probably 40+ N=1 apps. Multiply that across every function in every company.
2. The “we need a tool for this” → “let’s just build it” loop is now hours, not quarters. No procurement, no evaluation, no implementation consultants, no annual contract, no renewal negotiation.
3. SaaS vendors are about to face competition from software that didn’t exist as a category. If an N=1 tool costs $0 and 60 minutes to build and does exactly what you need, it’s hard for a $5K/year generic SaaS tool to compete — even if the generic tool is technically more polished. “Polished” stops mattering when the alternative is free and perfectly fitted.
4. The gap between “should this exist” and “does this exist” is collapsing. This is the real change. Software used to require a willing vendor or a willing engineering team. Now it just requires a willing human with Claude.
What Amelia’s app actually does
The attendee fills out if they need to park, what days, etc. The vibe’d app then matches the attendee to the right pass, pulls one of 4,000+ PDFs of the parking passes (each is unique!), and an email goes out looking like this:
Clean. Personal. Correct. For 4,000 people.
Seems simple, but really it’s not pre-AI. No software vendor would build this for us. We’re still waiting years for core feature gaps to be fixed in our events software vendor. Years.
And no intern burning a weekend. No mail-merge disaster at 11pm the night before. No “sorry, we sent you the wrong lot” apology chain. Just the right pass, to the right person, automatically.
The bigger picture
We have 20+ AI agents running SaaStr now. Artisan, Qualified, Agentforce, Monaco, QBee, 10K, Digital Jason — these are the named ones, the ones that run significant parts of the business end-to-end.
But the N=1 apps? The one-off internal tools that get built in an hour, used for a specific cycle, then updated or tossed the next time around? We’re going to have hundreds of these before long. And so will every other company that gets serious about this.
The app explosion hasn’t even started yet. What our Chief AI Officer Amelia built in a morning is the pattern, not the exception.
N=1 apps are so cool.
See you May 12–14 in San Mateo. Your parking pass will be in your inbox — sent by a 60-minute app.
