You should have a code from us. If you don't, this isn't quite ready for you yet — give us a few weeks.
We're building a smart collar that learns the specific things your specific dog does — not a list of generic behaviors picked by us. To get there, we need a small group of dogs and people willing to wear hardware, capture some video, and tell us what they see. In exchange, we're offering something we think is fair.
When our son was born we had two dogs. We were trying to introduce them to a baby while making sure rambunctious play stayed safe. The reality is that the dogs spent less time with us, not more — exactly the wrong direction.
Six months later we moved from Colorado to Baltimore. The new house had a fence Zoey was constantly escaping. So we spent $1,400 on a wireless fence to outline the physical fence we already had. At one point our two dogs had six different collars — training, bark, in-ground fence, dog-door, GPS, activity. Half the time they were missing. The other half the batteries were dead.
"What if you could have a collar that lets Elliot hang out with us in the kitchen — but warns him not to jump on the counter?"That question — the intersection of where the dog is and what the dog is doing — is what OneCollar is built around. We call it Blocation, which is a terrible name we'll probably keep. The hard part isn't building yet another tracker. The hard part is building a collar that recognizes the specific things your dog does, and lets you decide what to do about them.
That's the part we need help with.
Most "smart" pet products ship a fixed list of things they can detect. Walking. Running. Scratching. If your dog does something interesting that isn't on the list — tough.
OneCollar is built differently. The on-collar model doesn't try to name behaviors. It produces a continuous estimate of what your dog's body is doing right now — gait frequency, postural angle, rotational energy, vocalization intensity, and a learned residual that captures whatever isn't covered by those.
On top of that runs a separate library of behaviors. Each library entry is a pattern in motion-state space, plus a name, plus what to do about it. New behaviors get added to the library without retraining the collar.
This matters because you can teach the collar new behaviors. If your dog has a specific way of telling you they need to go out, or a particular jumping pattern that means "I'm about to steal something off the counter" — those become library entries. We capture synchronized video of your dog doing the thing, the platform learns the motion pattern, and from then on the collar flags it.
Novelty detection — the collar saying "I'm seeing something I don't recognize, want to label it?" — is the hardest part of all of this, and the most important. It's how the library grows.
Whatever we eventually charge for the behavior library — subscription, tier-gated features, per-behavior unlocks, none of the above — you don't pay it. Ever. Founders Pack accounts get every behavior, every update, every new library entry, for the lifetime of the platform.
When we ship the production collar (Rev 7 or whatever supersedes it), Founders Pack members get the first one free. Replacements and additional collars for other dogs in your house: founders pricing, well below retail, indefinitely.
Adam reads every email from this group. You'll have a real channel to flag bugs, request behaviors, push back on decisions, and tell us when something's broken. Not a support queue — a conversation.
The behaviors your dog helps us define get carried into the production library. Your data — anonymized, with your consent — is part of what makes the platform work for everyone else. We'll never sell it. You can pull out anytime.
// We'd rather promise too little here than too much. If we can't ship production hardware in a reasonable timeframe, we'll refund anything you've spent and your library access still stands. The library is the durable thing — the hardware is how we deliver it today.
The Rev 6 hardware lives on your dog the way any collar would. Sleeping, walking, playing, the whole day. Charge it overnight (~2x a week). If it falls off or breaks, tell us — we'll replace it.
Effort: zero, after setupThrough the app, you'll record 5–15 minutes of synchronized phone video while your dog does normal stuff — walking on leash, eating, getting up from a nap, jumping on the couch. The collar timestamps match the video. You tell us what's happening; the platform learns the motion patterns.
Effort: ~30–60 min/weekIf there's a specific thing your dog does that you want the collar to recognize — counter-surfing, the pre-bathroom circle, scratching at a specific door — film it a few times. Those become library entries that work for your dog first, and probably for other dogs too.
Effort: as much or as little as you wantThings will be broken. The app will crash. The collar will misclassify. A behavior we promised will be three months late. Honest feedback — even (especially) the harsh kind — is the only way this gets good. We'd rather hear "this is annoying" than nothing at all.
Effort: a few sentences, when something bugs youThe data you capture this week feeds the model that ships next month. Here's roughly what becomes available, and roughly when. Treat the timing as honest best-effort — we'd rather miss a date than ship something that doesn't work.
// As soon as a behavior is online, it's yours to use however you want. Pipe it into a webhook, get a push notification, see it in the timeline, train an automation — your call. We're not gating use behind the beta period.
Hierarchical inference, motion-state-plus-library, the silicon stack, and how the training data pipeline actually works. Written for technical beta users who want to understand what they're helping build.
Read the deep dive →We're applying for SBIR funding — a U.S. federal grant program that backs small companies building new technology. A short letter from a real beta user — generated from a few prompts in about three minutes — meaningfully strengthens the application. We'd be grateful.
Generate a letter →You've already got the access code, which means someone already pointed you here. Tell us a bit about you and your dog and we'll get back to you within a week.
If we can't fit you in this cohort, you'll be first in line for the next one.
// We don't have a sales team. We don't have a CRM. This form will literally email Adam. He'll write back personally. That's the deal for now.
Your default mail app should have opened with a draft to Adam — just hit send. If nothing happened, email founders@onecollar.ai directly with the same details. We'll be in touch within a week.