Magick Robotics · Seed round
Magick Guardian
The reward-only training and health system for big dogs.
Your dog learns in half a second. We built the system that never misses it.
Seed round · 2026 · info@magickrobotics.ai
The problem
Training fails on timing — and big dogs pay for it.
Humans miss the moment that teaches
Dogs link an action to its reward inside a window of about half a second. People almost always mark it late, so the lesson never lands.
A large, underserved segment
Roughly 20 million large-breed dogs in the US alone — the animals that most need reliable training and early health warning.
Problems caught too late
Big breeds face costly joint, gait and cardiac issues that owners typically notice long after the earliest signs appear.
No humane, reliable alternative
Shock and other aversive methods are being banned and abandoned — leaving owners without a dependable, welfare-first option.
The insight
A machine hits the marker window every rep — reward-only.
What a human can't do by hand, a device does every single time: mark the exact half-second your dog gets it right, and make sure the reward lands inside that window. One system, three jobs.
Train the dog
The perfectly-timed marker-and-reward loop, run flawlessly on every rep.
Coach the owner
Guardian Whisperer tells you the instant to mark and scores your timing.
Guard health
Activity, rest and gait tracked against your dog's own baseline for early warning.
The product
One system, a family of devices.
How it works
A closed loop the device runs flawlessly.
Cue
A spoken cue in your own voice — or nothing at all in capturing mode.
Mark ≤ 0.5s
A crisp tone fires inside the window, tagging the exact instant your dog got it right.
Reward in-window
A measured treat follows; the marker has already bridged the gap.
Adapt
Every rep is scored, and the plan raises or eases difficulty to match.
The moat: it coaches the human (Whisperer + Handler Score), transfers cues off-device so the dog obeys you, enforces welfare in software, and generates rep-level training data no one else has.
How the training works · 1 of 5
The moment that teaches.
Every rep is one closed loop, and the whole thing turns on a window about half a second wide. A device closes that loop the same way every single time.
- 1Cue — in your own voice, or nothing at all in capturing mode.
- 2Dog acts — the behaviour happens.
- 3Guardian detects — vision + motion, inside 0.5s.
- 4Mark — a crisp tone tags the exact instant.
- 5Reward — a measured treat lands in-window.
- 6Adapt — the rep is scored; difficulty adjusts.
A device hits that half-second window on every rep. A human — reaching for a clicker or a treat — almost always lands late, after the moment has already passed.
How the training works · 2 of 5
Everyone’s connected.
One quiet network runs the whole session: the app coaches you in-ear, the Guardian marks and cues the dog in your voice, and the dispenser drops the reward — all on time, every rep.
How the training works · 3 of 5
Live session — on your phone.
Sit · rep 14
The live training screen shows exactly what the Guardian saw and did — every reading the built app already reports, rep by rep.
-
Marker delay, in milliseconds
300 ms — inside the window — with timing quality 1.00.
-
Skill moving, live
Skill Δ +0.05, outcome comply, one measured treat logged.
-
Nothing to catch by hand
The device marks automatically; you just watch the rep count climb.
How the training works · 4 of 5
You’re getting better too.
Guardian Whisperer
Guardian Whisperer coaches the human in the loop — the moat no clicker has. Your timing tightens, your Handler Score climbs, and the cue transfers to you.
-
“Mark now” in your ear
The device knows the instant; it tells you, so you learn the feel of it.
-
Timing that visibly tightens
From ~0.9s late to ~0.4s across one week.
-
The cue transfers to you
Device-led → you cue → faded, so the dog obeys you.
How the training works · 5 of 5
Progress you can see.
Progress · 30 days
Because every rep is scored, progress is a picture, not a hunch — and the same sensors quietly watch health against your dog’s own baseline.
-
Per-skill reliability
Sit 92%, Recall 78%, Stay 61% — and recall climbed 62% → 95%.
-
Early health warning
Activity and rest tracked to baseline; a gentle flag when gait drifts.
-
A vet-shareable timeline
One tap turns a month of reps and readings into a record your vet can read.
The market
A large market, compounding fast.
Market sizes are company estimates drawn from third-party industry data; segment and penetration figures are company projections, not guarantees.
Business model
Hardware plus subscription — recurring by design.
A $100 refundable device deposit gets the Guardian on the collar — credited straight back as your first $100 of subscription.
- Guardian Whisperer$9.99/mo · free 3 mo
- Standard$19/mo
- Plus$29/mo
- Pro$39/mo
Roadmap
Each stage earns the next.
Where we are
Built and validated to spec — pre-launch.
Guardian Whisperer, built
A cross-platform app — iPhone, Android and web — built and validated to spec.
The loop, proven end-to-end
A high-fidelity training simulator validates the marker-and-reward loop to spec.
iOS · Android · web
The iOS build, Android and web are all passing.
Reward-only, enforced
There isn't even a command for an aversive — welfare is enforced in software.
Honest status: pre-launch. Hardware and the first pilot are next — no revenue or users are claimed.
Targets / projections
A credible path to ~$20M ARR in three years.
Targets and forward-looking projections based on current plans and assumptions; actual results will vary.
The ask
Raising a $5M seed.
To launch the MVP, prove training and health early-warning in real homes, and build Guardian Whisperer and the first hardware.
Forward-looking projections — not an offer of securities.