Project Overview

The Wildlife Rescue & Triage Agent (“Sage”) is an AI-powered conversational agent built in Voiceflow that combines receptionist and wildlife triage functions for the Wildlife Rehabilitation Center of Minnesota (WRC). Sage greets callers or web users, determines whether the Center is open or closed, and guides them through a structured intake process to classify the situation. Based on the animal’s species, condition, and risk factors, it provides species-specific triage advice and prepares the case for appropriate routing.

This project demonstrates how conversational AI can extend the reach of busy non-profits by offering accurate, structured, and empathetic first contact support 24/7, helping rescuers act quickly and safely until professional staff are available.

Genesis

The idea for Sage originated after a personal wildlife rescue experience. When a bat was found that appeared sick or injured, staff at the Wildlife Rehabilitation Center of Minnesota (WRC) were fully occupied and unavailable by phone. The urgency of the situation required independent action—safely containing the bat in a ventilated cardboard box and transporting it directly to WRC. That moment highlighted a gap: rescuers often need immediate, high-quality guidance when staff are unavailable.

Sage was conceived to fill that gap. By encoding triage logic, species-specific guidance, and WRC’s operational context into an AI agent, the goal was to create a digital first responder—always available, calm, and accurate—helping people make the right decisions when encountering wild animals.

Scope

• Agent workflow design (platform workflow): Built native workflows, variables, and agent architecture in Voiceflow to support intake, routing, and escalation logic across receptionist and triage functions.
• Prompt engineering (rules/spec authoring): Designed and refined instructions to control conversation flow, enforce safety filters, manage escalation rules, and maintain an empathetic, supportive persona.
• Information extraction (data ingestion/parsing): Created a structured species meta knowledge base covering Minnesota wildlife (songbirds, raptors, mammals, reptiles, amphibians) with intervention needs, referral flags, rabies risk, aggression, and care advice.
• Domain logic & modeling (business logic): Developed a variable framework to encode intake, routing, escalation, and emotional support states, enabling consistent downstream decision-making and case management.
• Presentation & UX: Implemented receptionist logic for open/closed hours handling, time-sensitive advice, and smooth transitions. Integrated persona design, safety-first phrasing, and clear triage question flow to balance empathy with precision.
• Quality & robustness (testing/ops): Iteratively tested flows, validated metadata handling, and enforced strict species normalization and escalation rules to ensure consistency, reliability, and safety in live use.

Skills Demonstrated

• AI/LLM: Prompt engineering, persona design, and safety-first conversation flow control.
• Conversational architecture: Multi-agent workflow design in Voiceflow with receptionist and triage functions.
• Data structuring: Creation of a structured species meta knowledge base with escalation, risk, and care advice fields.
• Logic & modeling: Variable framework for intake, routing, escalation, and emotional support handling.
• UX writing: Supportive, empathetic phrasing optimized for stressed or uncertain users.
• Branding & UI: Custom colors, chat imagery, and presentation design for a consistent identity.
• Ops & QA: Iterative testing, metadata validation, and enforcement of species normalization and escalation rules.

Deliverables

• Agent prompt & persona design: Full instructions for Sage, covering triage flow, receptionist logic, and safety rules.
• Species Meta Knowledge Base: Structured dataset of Minnesota wildlife with identification keywords, escalation rules, and care advice.
• Variable framework: System of variables to support intake, routing, escalation, emotional support, and case tracking.
• Conversation workflows: Voiceflow implementation of intake, routing, and escalation logic for end-to-end agent functionality.
• Custom branding & UI assets: Sage’s chat interface colors, imagery, and visual identity design.