Our path from working framework to production-grade platform
Operational Framework
The framework is already operational with Flask backend, OpenRouter AI, RAG knowledge retrieval, dual TTS engines (Edge-TTS and Coqui TTS), Live2D avatar, VTube Studio integration, YOLOv8 detection, and remote device support.
High Priority
Split app.py into route blueprints and service modules. Extract speech, detection, and avatar subsystems into dedicated service classes with clean interfaces. This refactor improves maintainability and enables independent testing of each subsystem.
High Priority
Align all route documentation with current endpoint names. The frontend already calls /chat and other modern endpoints, but some architectural diagrams still reference legacy /api/chat style labels. Update all diagrams to reflect the latest API surface.
Medium Priority
Add health check endpoints for all external services: OpenRouter, TTS providers, ffmpeg, VTube Studio, Web Avatar, and camera detection. Operators need quick status visibility before reception use to ensure all subsystems are online.
Medium Priority
Move detection and avatar settings from browser localStorage to a server-side config file. Add an admin UI for configuration management. Prevents configuration drift between sessions and enables centralized deployment management.
Medium Priority
Implement Cypress/Playwright-based frontend smoke tests covering chat flow, voice input, TTS, avatar sync, detection, and remote mode. The UI has many interaction paths that are expensive to verify manually across updates.
Low Priority
Build a dedicated staff interface for manual greeting triggers, info card display, avatar gesture control, and queue management without using the chatbot. Useful for staff who want direct operational control during peak hours.
Security audits, SSO integration, audit logging, RBAC, and compliance certifications for regulated industries. Deploy AiRA in healthcare, finance, and government environments with full confidence.
Cloud-hosted AiRA instances with tenant isolation, usage-based billing, and managed knowledge bases. Scale from a single kiosk to a nationwide deployment without infrastructure overhead.
Pluggable AI agents for specialized tasks: appointment booking, form processing, wayfinding, and integration with existing CRM/ERP systems. Extend AiRA with third-party and community-built agents.
