Research / 2026 Preprint

A Voice-First Agentic AI Architecture for Veteran Navigation, Transition, and Personalized Support

A human-centered framework demonstrated through the ValorBuddy platform

Eugene Ezenwa Ebem

Co-Founder & Chief Technology Officer, Tagus Technologies LLC · Dallas-Fort Worth, Texas, USA

PUBLICATIONPreprint / Working Paper
DATESeptember 2026
DOI10.5281/zenodo.25881294
Read full-text PDFView DOI recordAuthor ORCID

Abstract / Research Summary

Veterans and transitioning service members often interact with a fragmented ecosystem of benefits information, health resources, employment services, appointments, community organizations, family-support programs, and civilian-transition resources. This paper proposes a voice-first agentic artificial intelligence architecture designed to reduce navigation burden through personalized, conversational assistance. Demonstrated through ValorBuddy, the architecture follows a continuous assistance pattern: Know Me → Guide Me → Connect Me → Help Me Act → Grow With Me. It presents the architectural model, agent orchestration approach, consent-based personalization strategy, security and responsible-AI controls, and an evaluation framework.

This manuscript describes an architecture and evaluation framework. It does not claim clinical efficacy, VA endorsement, benefits-adjudication authority, or validated outcome improvements.