TailTracker

The problem

Shelters run on memory and whiteboards.

Enterprise shelter software like PetPoint exists, but it was built for data entry clerks at a desk — not for kennel workers standing in front of a dog with a hose in one hand and twenty animals left to check. Care logging is slow, inconsistent, and often skipped entirely. There's no reliable way to know, at 6am, what actually happened overnight.

The result: animals fall through the gaps. Feedings get missed. Medication timing drifts. The morning shift inherits chaos instead of context.

The solution

Make the right thing the fast thing.

TailTracker is a full shelter management system built around one constraint: the person logging care is standing in a kennel, probably wearing gloves, and has fifteen seconds. QR code on the kennel. Scan. One tap to log fed, medicated, cleaned, or flagged. Done. Move on.

Everything else — vet triage, AI briefings, adoption compliance, cruelty case tracking, grant pipeline — is built on top of that core loop. The system serves 17 distinct roles, from kennel staff to executive director, each with a tailored view of exactly what they need to know right now.

Role-specific AI briefings fire three times a day so every shift — morning, afternoon, evening — can start informed instead of guessing. The vet hub ranks animals by medical urgency. The lost & found module uses AI to match incoming strays against open reports. The social tool generates adoption bios from a photo.

Validation

Stress-tested by a fleet of AI agents.

To test the system the way a real shelter would stress it, I generated a full year of operations — 14,634 intakes, nearly 4 million care logs, 12,508 adoptions, and a stocked supply room — then turned a fleet of nine AI agents loose on the live application, blind to each other, each playing a role: kennel and ranch workers logging care, vets dispensing medication and clearing cases, counselors processing adoptions, intake staff admitting strays, social media drafting captions, purchasing restocking shelves. Across 30 cycles of concurrent operation they ran 2,580 real actions through the genuine endpoints — with zero failures, every write provably isolated from production. The run also drove the full supply economy — food, veterinary meds and vaccines, and ranch feed — through its whole cycle: consumption depleted stock, the low-stock alert fired, and purchasing replenished it. A second fleet of role-specific QA agents then audited every domain, passing 45 of 50 checks — the rest flagging deliberately-deferred features, not defects.

Status

Built with the shelter.

TailTracker was built from the start in close collaboration with shelter leadership — every workflow, every role, every constraint came from conversations with the people who will actually use it. The core care-logging loop and role-based system are in place; next steps, rollout cadence, and what to refine are being worked through in upcoming partner reviews. If you'd like to see it in action, reach out directly.