Your platform. Your data. Your AI.
The creator economy is a $250 billion market. Big Tech takes 20% — then taxes the data, then sells it back. Every OnlyFans payout, every Fansly subscriber, every fan message is processed by infrastructure that has no skin in your game and every incentive to harvest what you build. GenieHelper flips the stack. Your data stays on a server you subscribe to. Your AI runs on models we host — uncensored, unfiltered, ungoverned by corporate content policy. Your fan relationships, your content performance, your posting schedule, your platform credentials — encrypted at rest, never leaving our infrastructure, never feeding someone else’s training set.Quick Start
Set up your account and connect your first platform in minutes
Platform Overview
Explore all features: fan CRM, content scheduling, media pipeline, analytics
AI & Memory System
Learn how GenieHelper’s private AI and memory stack works
Pricing & Tiers
Starter, Creator, Pro, and Studio plans — from 499/mo
What GenieHelper does
Multi-Platform Management
Connect OnlyFans, Fansly, Instagram, and TikTok from one interface. Scrape stats, track engagement, and manage sessions without switching tabs.
Private AI Chat
Dolphin 3 8B and Qwen 2.5 run locally — zero OpenAI dependency. Your conversations, your models, your server.
Fan Intelligence CRM
Per-fan profiles with 71 engagement fields, purchase history, message sentiment, and AI-suggested follow-up tasks.
AI Content Drafting
Uncensored Ollama models write captions, fan messages, and post concepts in your voice — trained on your persona, not a generic template.
Post Scheduling
AI-suggested editorial calendar with tier-enforced queue limits and BullMQ-powered reliability.
Media Pipeline
Watermarking, teaser clipping, thumbnail generation, and metadata stripping — all local FFmpeg, no cloud uploads.
Steganographic Watermarking
Every piece of purchased media gets an invisible, fan-specific watermark. If a file leaks, you know exactly which fan’s purchase it came from.
Automation Triggers
Rule-based automation chains fire on fan behavior: win-back on cancel, upsell after PPV, birthday greetings, re-engagement for inactive fans.
Infrastructure philosophy
GenieHelper is not a SaaS play built on managed cloud services. It runs on a dedicated server you subscribe to — IONOS dedicated, Ubuntu 24, 16GB RAM, 1TB NVMe, fully self-hosted.No cloud. No vendor lock-in. No corporate content moderation. All AI inference runs locally via Ollama. Nothing leaves the VPS.
| Layer | Technology |
|---|---|
| LLM inference | Ollama — Dolphin3 8B, dolphin-mistral 7B, Qwen 2.5 |
| Agent orchestration | AnythingLLM (fork) — multi-user, MCP auto-boot |
| API server | Express + Prisma (port 3001) |
| CMS / data layer | Directus 11 (port 8055) |
| Primary database | PostgreSQL 16 with pgvector + AGE graph extension |
| Job queue | BullMQ + Redis (concurrency:1) |
| Browser automation | Stagehand (Playwright, port 3002) |
| Frontend | React 18 + Vite + Tailwind (port 3100) |
| Credential encryption | AES-256-GCM per-user vault |
Subscription tiers
| Tier | Price | Audience |
|---|---|---|
| Starter | $0/mo | Try the platform |
| Creator | $49/mo | Solo content creators |
| Pro | $149/mo | Power users with advanced fan CRM |
| Studio | $499/mo | Agencies and multi-account operations |
For developers
GenieHelper is built on open-source primitives: AnythingLLM, Directus, Ollama, Playwright, BullMQ, DuckDB, and PostgreSQL with pgvector and AGE. The interesting engineering is in the agent intelligence layer — specifically how the system builds, retrieves, and promotes knowledge over time.System Architecture
Web architecture, data flow, and middleware stack diagrams
MCP Tools
83 tools across 6 plugins: cms.directus, ai.ollama, web.stagehand, and more
Memory & Retrieval
HyDE, RRF, synaptic propagation, Shannon entropy gating, Hebbian consolidation
API Reference
REST endpoints for chat, queue management, credentials, and skills