Coterra · Loading
PRIVACY & AI · COTERRA
We use AI for two narrow jobs — reading the airline confirmations you forward, and drafting a starting packing list. The AI never knows who you are, when your kids were born, where you live, or anything else identifying. And we never train AI models on what you give us.
WHY WE BUILT IT THIS WAY
One of the parents we built Coterra for is the one who closes a browser tab the moment a site asks for medical history 'to personalize your experience.' Sean's wife is one of those parents. So is yours, probably.
That parent doesn't oppose technology. She just refuses to be the product. She'll happily use a tool that does specific work for her — as long as the tool is structurally unable to abuse it.
Coterra is built around that test. Not "we promise we won't" — "we built it so we can't."
WHAT WE USE AI FOR
AI is in the background, doing work you'd otherwise have to do yourself. You never see a chat box, never type a prompt, never read a 'powered by AI' label.
1 · READING EMAILS
When you forward an airline confirmation to your trip inbox, we use AI to pull out flight numbers, times, airports, and passenger names from the raw email and convert them into a clean record. We don't analyze, summarize, or learn from your trip.
The provider never retains the email after processing — we verify this with each AI provider via their no-retention API setting. The parsed result stays in your Coterra account.
2 · DRAFTING A PACKING LIST
When a trip is ready, we draft a packing list. We send the AI a categorical version of your trip: 'A family of 2 adults + 1 child (14 months) + 1 child (3 years) is taking a 7-day trip to a temperate destination in late summer, returning to the US from Canada.' That's the whole prompt. No names. No birthdays. No address.
The AI replies with a categorized starting list — diapers in the right size, age-appropriate snacks, sun protection appropriate to the climate. Then Coterra's own code (not AI) adds the personal touches: EpiPen if your kid has a peanut allergy, ibuprofen with the right measuring tool, customs-restricted items removed.
The personal information never leaves our database to do this step. There's no AI in it at all.
WHAT WE DO NOT USE AI FOR
Customs and TSA rules. Hand-maintained. Every answer cites the actual CBP, USDA, or TSA rule with a 'last verified' date. AI would be too fast and too wrong; people pay fines for our mistakes.
The emergency translation card. The phrases on the card ("My child needs a pediatrician," "My child has [allergy]") are hand-translated per destination language, reviewed by a native speaker. No machine translation. The wife test would fail one in an actual emergency.
Embassy and pediatric-hospital contacts. Hand-researched per destination, phone numbers verified annually.
The pattern: when accuracy matters more than speed, we hand-curate. When fluency at scale matters more than personalization, we use AI. Both promises are kept by what we choose not to send, not by what we choose to send.
THE EXACT INPUTS
What the AI does see
What the AI never sees
THREE PRIVACY PROMISES
PROMISE 1
Your emails, trips, and family info never become someone else's autocomplete. Locked in our Privacy Policy and in every provider contract we hold.
PROMISE 2
Even when AI helps draft a packing list, it never knows your name, your kid's birthday, your allergies by name, or your home address. The list above is the entire visible context.
PROMISE 3
With a passphrase you choose. We can't recover them if you forget the passphrase — that's the trade-off for keeping them private from us. Ships at v1.5 when document scanning launches.
THIS PAGE IS A LIVING DOCUMENT
Every AI feature Coterra ships must keep this page synchronized with the actual code. That's not a promise; it's a process gate. Any pull request that adds or changes an AI call has to update this page in the same change, or the request gets bounced before it merges.
If you ever see this page describe behavior that does not match what Coterra actually does, tell us. We will fix it within 24 hours and acknowledge the discrepancy publicly.