yaowu-ai/aida

AIDA is a full stack web application for training sales and customer service staff using AI. It lets a company build practice scenarios where a trainee talks to an AI customer, either by voice or by text, and then receives a detailed scorecard on how they did. It is built with Next.js, React, Prisma, MySQL, and Redis, and it uses VolcEngine's real-time communication, speech recognition, and text-to-speech services to power the voice conversations. The product walks through four steps that form a loop. First, a training lead can drop in a plain sales playbook document and the built-in AI automatically extracts a scene, a customer persona, and a scoring rubric, turning it into a ready-to-practice scenario without any manual scripting. Second, a sales rep practices against that AI customer, who behaves like a real, sometimes difficult buyer who pushes back on price or asks pointed questions, available any time with no risk to real customers. Third, each session produces a scorecard with an overall score, ratings on specific skills like needs discovery or objection handling, turn by turn coaching notes with model answers, and a score trend over time. Fourth, an HR or organization admin can bulk import their team structure, assign quotas for voice and text minutes and storage, and track usage on a dashboard. Under the hood, the app runs as a single Next.js service backed by a MySQL database through Prisma, with background jobs handled by BullMQ and Redis for things like batch evaluation and a cron scheduler. It also includes JWT based authentication with phone number SMS login, object storage through VolcEngine, and a super admin only flow for granting organizations more training minutes or storage quota, usable either through a web page or a command line script for batch operations. This project is aimed at companies or teams that want to run structured, measurable sales or customer service training internally, and at developers who want to see a real example of an AI voice training platform built on Next.js.