sethyu1/errowsai

Errows is a full-stack platform where users create custom AI characters, each with its own personality, appearance, and backstory, then interact with them through text chat, real-time voice calls, and generated images or videos. Users can also share their characters and posts with other users in a community feed, similar to a social network built around these AI personas. Character creation walks a user through choices like gender, visual style, and personality, with AI assistance from xAI's Grok model helping refine the persona, plus automatically generated avatars and greeting messages. Conversations stream in as the AI responds, and the platform remembers context across a session. Voice calls work through a real-time audio service combined with a speech-to-text and text-to-speech pipeline, so a character can hold a live spoken conversation rather than just typed messages. Users can also request custom images and videos of their characters, generated asynchronously in the background. The business side runs on a coin-based economy: users buy coins and subscriptions through Stripe, redeem gift codes, and complete daily tasks for rewards, which unlock features like extra media generation. There's also a full admin console for managing members, moderating content, tuning AI generation settings, and viewing usage and revenue statistics. Technically, this is a monorepo managed with pnpm, containing a React web app, a Capacitor-wrapped mobile app, a separate admin console, and a backend built from independent microservices using the Moleculer framework, all backed by a PostgreSQL database. Getting it running requires Node.js 18 or newer, PostgreSQL, and a long list of API keys for services like Stripe, AWS S3, Agora for real-time voice, MiniMax for text-to-speech, and xAI, so this is not something to spin up casually. It comes with deployment scripts for pushing builds to a server over SSH and rolling them back if needed. This is a mature, production-oriented codebase best suited for developers who already understand microservices architecture and want to study or build a similar AI companion product. It is released under the MIT license.