Nexora AI
AI workforce platform automating customer communications, lead management, and business workflows across WhatsApp, Instagram, and Messenger.
Type
Tech stack
Connected services
Documentation
Full
Setup, architecture, and how the pieces fit together.
Maintenance
Active
Actively maintained with ongoing updates.
Description
NEXORA AI v2
Multi-tenant WhatsApp/Instagram/Messenger AI automation platform for Indian SMBs - one backend, one database, unlimited client businesses. See ARCHITECTURE.md for how it fits together.
Stack
- Backend: FastAPI (Python 3.12), deployed on Render
- Frontend: Next.js 14 + TypeScript + Tailwind, deployed on Vercel
- Database: PostgreSQL + pgvector (Supabase)
- Cache/Queue: Redis + ARQ (Upstash)
- AI: Groq LLaMA + OpenAI embeddings
- Storage: Cloudflare R2
- Messaging: WhatsApp Cloud API, Instagram Graph API, Messenger
- Auth: JWT + bcrypt, Google Sign-In
Getting started
The whole app in one command, no Supabase/Upstash/Meta credentials needed:
See CONTRIBUTING.md for that quickstart's bootstrap steps,
native (hot-reload) local setup, and the branching/promotion workflow
(staging → main).
Testing
CI (.github/workflows/) runs both on every pull
request and on every push to main/staging. See
CONTRIBUTING.md for the integration tier
(needs real Postgres + Redis) and the E2E suite.
Non-obvious engineering
A quick scan reads this as "another FastAPI+Next.js chatbot SaaS." The parts below are where that description stops being the whole story — specific design decisions made under real production constraints, not generic framework usage:
FAQ
- What accounts and services do I need to set up before running this?
- You need Supabase for PostgreSQL and pgvector, Upstash for Redis, OpenAI for embeddings, Groq for the LLaMA model, and Cloudflare R2 for file storage. You also need credentials for WhatsApp Cloud API, Instagram Graph API, and Messenger if you want to connect those channels. Google Cloud and Sentry are optional for auth and error tracking. All of these have free or trial tiers sufficient to get the platform running locally.
- What is the tech stack?
- Backend is FastAPI on Python 3.12, frontend is Next.js 14 with TypeScript and Tailwind. The database is PostgreSQL with pgvector extensions on Supabase, and Redis on Upstash handles caching and job queues via ARQ. AI runs on Groq LLaMA with OpenAI embeddings. If you know FastAPI and Next.js, the architecture will be straightforward.
- How do I get it running locally?
- CONTRIBUTING.md has a one-command quickstart that runs the whole app without needing Supabase, Upstash, or Meta credentials upfront. It also covers native hot-reload setup for development and the staging-to-main promotion workflow.
- What testing and CI is included?
- GitHub Actions runs on every pull request and push to main or staging. CONTRIBUTING.md documents the integration tier (requires real Postgres and Redis) and the E2E suite so you can verify the platform end-to-end before deploying.
- What documentation is included?
- Full documentation covers setup, architecture (ARCHITECTURE.md), contribution workflow, and testing. The codebase is actively maintained.
Domains
Seller
Reviews
No reviews yet.
Related reading
A breakdown of what an online store for the Egyptian market actually needs, and where starting from a finished build saves you a quarter.
