StaffPilot — AI-Powered Social Media Management SaaS by Abdulqoyyum AileruStaffPilot — AI-Powered Social Media Management SaaS by Abdulqoyyum Aileru

StaffPilot — AI-Powered Social Media Management SaaS

Abdulqoyyum Aileru

Abdulqoyyum Aileru

StaffPilot is a full-stack, multi-tenant SaaS platform that helps marketing teams plan, generate, schedule, and publish AI-crafted content across every major social network — from a single dashboard.
What it does
Marketing teams connect their Facebook, Instagram, LinkedIn, Twitter/X, and TikTok accounts via OAuth 2.0, then let StaffPilot handle the entire content pipeline: AI-generated text (Gemini 2.5 Flash, GPT-4, or Claude), AI-generated images (Gemini Imagen 4), and short-form video (Google Veo 3). Posts are queued and published automatically by a Celery Beat scheduler. Campaigns can be managed and tracked across Google Ads and Meta Ads from the same interface.
A RAG (Retrieval-Augmented Generation) pipeline lets teams upload internal PDFs and DOCX files — product docs, brand guides, tone-of-voice documents — so every AI-generated post is grounded in company-specific facts rather than generic output. Documents are parsed, chunked, embedded with Gemini Embedding 001, and stored in a Pinecone vector database namespaced per tenant.
The platform also ships with configurable AI chat assistants (Digital Marketer, Customer Support), SEO keyword research via SerpAPI, a media storage layer backed by Cloudinary, and Stripe-powered subscription billing.
Architecture & Tech Stack
The backend is a FastAPI application (Python 3.11, async SQLAlchemy 2.0, PostgreSQL 15) deployed as three Railway services: API server, Celery worker, and Celery Beat scheduler. Redis 7 serves as both the task broker and cache layer. All long-running operations — content generation, document ingestion, scheduled publishing, email notifications — are handled asynchronously by Celery workers.
The frontend is a Next.js 15 (App Router) application built with React 19, TailwindCSS 4, Radix UI primitives, Recharts for analytics, and Framer Motion for transitions. It is deployed on Vercel.
The LLM layer uses a factory pattern, making it trivial to switch or extend AI providers (Gemini, OpenAI, or Anthropic) through a shared interface without touching any calling code.
Multi-tenancy is enforced at the database query level: every JWT carries a tenant_id, and every query is automatically scoped to that tenant's data — ensuring full isolation between organizations on the shared infrastructure.
Key integrations: Gemini (text/image/video/embeddings), OpenAI, Anthropic, Pinecone, Cloudinary, Stripe, SerpAPI, Facebook Graph API, Instagram Graph API, LinkedIn API, Twitter/X API (OAuth 1.0a + 2.0), TikTok API, Google Ads API, Meta Ads API.
Like this project

Posted Jul 6, 2026

Built a production SaaS with AI content generation, 5-platform publishing, RAG knowledge base, async scheduling, and multi-tenant billing