Truth Computing
01 / Problem

Not running out of people. Running out of time.

~6.3M animals enter U.S. shelters each year; on the order of 600,000 are euthanized — often for time and space, not a lack of willing homes.

The animals in our shelter system are not running out of people who would love them. They are running out of time before the right match is made. Shelters operate in isolation, with no shared infrastructure to surface a compatible adopter across facilities.

  • Behavioral intake is inconsistent. Temperament, history, and compatibility live in free-text notes or one staffer's head.
  • Matching is passive. Adopters scroll listings; shelters can't proactively surface the right adopter for an animal. Good matches depend on luck.
  • Length of stay is the silent killer. The longer an animal stays, the worse its outcomes — and staff often can't see who is most at risk until decisions are forced.
  • Shelters can't see each other. A full facility has no easy way to find a partner with capacity and demand.
  • Returns are costly. Poorly matched adoptions come back, consuming scarce time and space.

Existing shelter-management systems handle records and listings well. Hazel doesn't replace them — the gap is in behavioral standardization, active matching, length-of-stay risk, and cross-shelter coordination.

02 / Goals & Non-Goals

What V1 commits to — and the line it will not cross

Goals

  • Capture a standardized behavioral profile that travels with every animal.
  • Actively match animals to compatible adopters by lifestyle and household fit.
  • Surface length-of-stay risk early so staff can prioritize and intervene.
  • Coordinate cross-shelter transfers to reach capacity and demand elsewhere.
  • Measurably shorten length of stay, raise live-release rate, and reduce returns.

Non-Goals (V1)

  • Does not replace shelter-management / records systems — it integrates.
  • Makes no euthanasia or life-or-death decision, ever. It informs human judgment.
  • Does not handle clinical/veterinary record-keeping beyond placement-relevant fields.
  • No payments, donations, or adoption checkout in V1.
03 / Users

Built for the kennel floor

Volunteer-heavy, mixed-tech, under-resourced. If it's slower than a sticky note, it dies.

Intake Staff

Fast, structured behavioral intake on a phone or tablet at the kennel; consistent fields every time.

Behavior / Enrichment Staff

Update temperament and behavior notes; flag training needs; see at-risk animals.

Adoption Counselor

A ranked list of compatible adopters for an animal — with the reasons for each match.

Shelter Manager

Length-of-stay risk view; capacity and outcome dashboards; transfer opportunities.

Transfer / Foster Coordinator

See partner-network capacity and demand; match at-risk animals to fosters and partners.

Adopter (indirect in V1)

A profile capturing home and lifestyle so they're matched well and the adoption sticks.

04 / V1 Scope

Five modules, built and rolled out together

Full-platform V1. Each module is validated with the design partner before it expands across a transfer network.

Module 01

Behavioral Intake Profile

Standardized, mobile-first assessment at intake — temperament, energy, history, compatibility with kids / cats / dogs, handling and training flags. One consistent schema, captured once, feeding everything downstream.

Module 02

Adopter Matching Engine

Compatibility scoring between animal profiles and adopter profiles, surfaced both directions. Every match is explainable — it shows why. The engine ranks and reasons; staff decide.

Module 03

Length-of-Stay Risk

Track length of stay and surface a prioritized view of animals most at risk by time, space, and placement difficulty — so behavior, foster, marketing, or transfer intervention happens before a decision is forced.

Module 04

Cross-Shelter Transfer

A network view of partner shelters', rescues', and fosters' capacity and demand. Match an at-risk animal to space and likely adopters elsewhere, with the profile attached and outcomes tracked.

Module 05

Public Adoption Profiles & Sync

Auto-generated, high-quality public listings from the same behavioral profile — no double entry. Sync to the shelter site and listing platforms where integrations allow; always current.

Cross-cutting

Human-in-the-Loop, Always

Hazel ranks, reasons, and flags; staff decide. No automated life-or-death logic, ever. Mobile-first and kennel-side; integrates with PetPoint / Shelterluv / Chameleon-class systems; Spanish support.

05 / Success Metrics

An honest scoreboard, starting at zero

North-star: length of stay to positive outcome.

Median length of stay

The core lever — less time in shelter, better outcomes.

Live-release rate

The outcome that ultimately matters.

Return rate (post-adoption)

Did better matching make adoptions stick?

Time-to-adoption, at-risk animals

Are early flags plus intervention working?

Transfers resulting in placement

Is cross-shelter coordination real?

Euthanasia for time/space reasons

The number we exist to drive down.

06 / Rollout

Go deep with one partner, prove it, replicate

The same playbook behind Feynman: find the real workflow gap, build a validated tool, measure, then point it at the next shelter. ~3,500 U.S. shelters run today without shared infrastructure.

Phase 1

Discovery

Identify shelter partners and learn how intake and placement actually operate — before writing a line of code.

Phase 2

Design-partner build

Build V1 with one shelter or pound. Start with standardized behavioral intake feeding active matching; expand as each module is validated.

Phase 3

Pilot & measure

Run with real animals and real adopters. Publish the scoreboard. Fix what the numbers and staff tell us.

Phase 4

Transparent report

What we set out to do, what the numbers say, what changed, and what we got wrong.

Phase 5

Replicate

Extend to a transfer network of nearby shelters, contingent on resources and mutual agreement on terms.

07 / Risks & Open Questions

What could go wrong, said plainly

  • Adoption by busy, volunteer-heavy staff. If it's slower than a sticky note, it dies. Mitigation: build with staff; kennel-side mobile; faster than today on day one.
  • Behavioral assessment validity. Temperament assessment is contested science; over-claiming would be harmful. Mitigation: conservative, established frameworks; present uncertainty honestly.
  • The euthanasia line. Hazel must never appear to automate or recommend a life-or-death decision. Mitigation: hard product boundary — flag risk, never decide.
  • Integration constraints. Existing shelter systems vary in openness. Open: what system does the partner run?
  • Network effects. Transfers need more than one shelter. Mitigation: deliver standalone value first; transfers compound as the network grows.

Open questions: Which shelter or pound is the first design partner, and what management system do they run? What behavioral assessment framework do they trust? Where is the biggest measured loss — intake quality, matching, or length-of-stay visibility? Which public-listing channels matter most?

See the shelter data & initiative →