Why Outpost exists.

Every camping app I tried wanted to be a booking platform. None of them wanted to be a trip planner.

The category is dominated by two shapes. The first is the campground directory — Recreation.gov, Hipcamp, The Dyrt — built around where to go. They're great for that. They're useless for what to bring and how to get the trip together. The second is the generic packing-list app — generic checklists, drag and reorder, share with a partner. They work, technically. They're also dead-paper: someone built them in 2014 and the world moved on.

The gap is the middle. The work of planning a 3-day trip with one or two friends and a vehicle: the itinerary, the meals you'll cook there, the gear you actually need for this particular style of trip, who's bringing what. That's a real twenty-tab notebook problem and nobody had a single coherent surface for it.

A weekend trip a few weeks ago. Three friends. Four kinds of camping in one trip.

We did a thing where the first night was car camping at a state park (full setup — tent, cooler, big chairs), the second night was a moderate backpack into a wilderness area (one cookset, light tarp, water filter), and the last morning we wanted to drive somewhere with cell signal to work for a couple hours before the drive home. The packing list for night 1 and the packing list for night 2 had a 30% overlap and 70% conflict.

I built three lists in Apple Reminders. They were a mess by Friday afternoon. I dropped a meal plan into Notes. I didn't have it on my phone in any usable way at the trailhead. We forgot the camp soap, brought two extra stoves, and someone left a $90 sleeping pad in the car at the trailhead because it wasn't on the "what we're carrying tonight" list.

I came home and started Outpost on Monday.

The thing the app does that nothing else does is the smart packing checklist.

Pick a camp style — tent, backcountry, car camping, RV, cabin. Pick the length, the season, who's coming. The app generates a starting checklist tuned to that combination, not a generic template you have to delete half of. Backcountry doesn't show "ice chest." RV doesn't show "water filter." Cabin doesn't show "tent stakes."

That's the part that took the most engineering and isn't the part most users will notice. Underneath, it's a small rule system over a gear catalog — a few hundred items, each tagged with which camp styles it's relevant for and what conditions trigger it. You see a clean checklist; the system has done the filtering.

And if the trip has multiple legs (like the one a few weeks ago), each leg gets its own packing context. The car-camping list and the backcountry list both exist on the same trip. They share what's shared and don't pretend to share what isn't.

The Discover tab uses Foundation Models on the phone. Not a cloud LLM in the loop, ever.

The Discover tab is where you go when you don't know where to go. Tell it "long weekend, within four hours of KCGI, want some hiking but mostly stargazing, dog-friendly." It runs the prompt against Apple's on-device Foundation Models, cross-references the catalog of state parks and the live NPS / Recreation.gov data, and gives you a few candidates with quick reasoning.

The reason that has to be on-device — and not, say, GPT-via-server — is the same reason tmpo's data never leaves the phone. The standing rule is: if the AI feature requires sending your stuff to my server (or someone else's), the app doesn't have that feature. With Foundation Models, suddenly it doesn't have to. That changes which features are even possible while keeping the architecture honest.

A short list of things I cut on purpose.

• Booking flow. Recreation.gov already exists. Outpost links out, doesn't reimplement the reservation flow.
• Social. No feed, no "follow Ethan's trips," no public trip log. Trip data syncs through your private iCloud, not mine.
• Gear-store affiliate links. The catalog points at gear types, not specific products to drive REI revenue.
• Group-trip realtime sync via the cloud. v1 syncs through CloudKit per user. Sharing a trip with someone means exporting it to them; in v2 I'll look at proper collaborative editing.
• Weather forecasting. That's Weather.Co's job. Outpost surfaces the relevant slice (current + forecast for the trip dates) and links out for deeper.

The v1 footprint is intentional. Six months from now I'll know which of these was a mistake. Right now the goal is the smallest thing that does the actual job — pack the right stuff, get the trip together, find a good place when you don't have one.