AI Features

The AI Features page lets you enable and configure PawPlacer's artificial intelligence capabilities. Each feature is independent; you can turn on just the ones that fit your workflow.

Accessing AI Features

1. Open Settings
2. In Integrations, click AI Features

This page requires Admin permissions.

Privacy and Data Handling

A privacy notice appears at the top of the page explaining how your data is handled. Key points:

• OpenAI states that API data is not used to train its models by default. Limited retention for abuse monitoring may apply unless a Zero Data Retention arrangement is in place
• Matching filters direct identifiers and blocks or redacts PII-like fields and values where possible before data is sent
• No AI features are enabled by default; you choose exactly what to turn on
• AI output is assistive only; staff must review it before making placement or operational decisions
• Links to the Privacy Policy and Terms of Service are provided in the banner

Regional Restrictions

AI Matching is not available in countries covered by GDPR regulations. If your organization is located in a restricted region, the AI Matching toggle will be disabled and a notice will explain why. AI Descriptions and AI Data Mapping are available in all regions.

AI Matching {% badge text="Beta" /%}

AI Matching goes beyond simple keyword filters. It reads the full context of each applicant and pet profile to surface compatibility insights your team might otherwise miss.

What It Does

• Provides quick compatibility estimates on every applicant profile once matching profiles are generated
• Produces calibrated detailed scores with factor breakdowns, confidence, readiness, missing-evidence guidance, strengths, concerns, and plain-English explanations
• Lets staff compare two or three candidates side by side
• Learns conservatively from staff feedback, placement progress, and confirmed outcomes
• Considers household setup, lifestyle, experience, and what each applicant is actually looking for
• All results are reviewed by your team before any decisions are made; AI Matching is an advisory tool, not a decision-maker

Enabling AI Matching

Toggle Enable AI Matching to on. Once enabled, PawPlacer builds the initial matching index and queues profile refreshes when relevant records or enabled fields change. You'll also see several additional options:

• Preview compatible fields: Opens a dialog where you can review eligible form fields, deselect anything you do not want included, and enable the selection. PII-like fields are labeled and excluded or disabled.
• Organization learning: Shows learning-signal totals, confirmed placements, average confidence, top-three placement success, and learned factor priorities.
• Reset learning: Removes staff feedback and learned weighting while preserving generated profiles and current matches. A confirmation dialog prevents accidental resets.
• Clear matching data: Removes generated profiles, embeddings, and cached match results without deleting pets, adopters, or fosters. Use it when you need to rebuild from a clean state.

Data Sent for Matching

The page shows exactly what data is sent for each match. Two preview boxes display sample data:

• Pet data: species, breed, size, age, temperament, health status, description, adoption status, and any custom fields you've marked for matching
• Person data: household setup, lifestyle, experience, preferences, and any custom fields from your application forms marked for matching

How AI Matching Works

An expandable FAQ section in the AI Matching area covers three levels of detail:

• How does this work?: A plain-language overview of the matching logic
• How does it actually work?: Explains the two-tier system: quick estimates (fast, embedding-based) and full AI-scored matches (detailed, LLM-powered)
• Engine room: Technical details covering embeddings, HNSW indexes, LLM scoring, the write path, and PII defenses

Reviewing and Improving Matches

Detailed match cards separate compatibility from workflow readiness. Staff can review five possible factors—preferences, household, experience, schedule, and care needs—along with confidence and missing profile details. Missing factors remain unknown instead of being assigned a neutral score.

Use Compare to select two or three results and review scores, readiness, factors, strengths, concerns, and the best next profile update side by side.

Staff can mark a match as helpful or not quite right. Negative ratings can include a reason category and optional note. The Progress menu records funnel events such as shortlisting, meet-and-greets, trial placements, completed fosters, declines, and withdrawals. PawPlacer combines these signals with confirmed outcomes using sample safeguards, so early feedback makes only small adjustments.

AI Descriptions

AI Descriptions generates written pet bios using data from each pet's profile. This can help you create consistent, engaging descriptions without writing each one manually.

Enabling AI Descriptions

Toggle Enable AI Descriptions to on.

A data preview box shows the type of information used to generate descriptions: species, breed, age, size, temperament, and health status. No data is sent unless you explicitly enable this feature.

Once enabled, AI description options become available on individual pet profiles.

AI Data Mapping

AI Data Mapping uses AI to help map imported data fields to PawPlacer fields during bulk data uploads. This speeds up the import process when column names in your spreadsheet don't exactly match PawPlacer's field names.

Enabling AI Data Mapping

Toggle Enable AI Data Mapping to on.

A data preview box shows an example of how form field names and values are mapped.

Permissions

Only team members with Admin access can view or modify AI feature settings.