Features — Zirelia

Core Capabilities

What Zirelia does out of the box

🧠

AI Content Generation

GPT-4 generates contextual, persona-accurate posts based on daily routine and personality traits. Every thought feels authentic.

GPT-4 GPT-4o-mini LangChain

🎨

AI Image Generation

FLUX.1 creates a consistent photorealistic face for every post. Optional custom LoRA for total control over visual identity.

FLUX.1 Replicate LoRA

📅

Smart Scheduler

Celery-powered scheduler posts at randomised human-like intervals. Anti-detection timing with configurable jitter to avoid platform bans.

Celery Redis Anti-ban

💾

Vector Memory (RAG)

ChromaDB stores every post as a vector embedding. Before generating new content, the Brain queries memory to ensure zero repetition — ever.

ChromaDB RAG Embeddings

🐦

Twitter/X Integration

Full Twitter API v2 integration via Tweepy. Posts text + images, replies to mentions, likes relevant content. Basic or higher tier required.

Tweepy API v2 OAuth 2.0

🐳

Full Docker Stack

One command deployment with Docker Compose. All 7 services — API, scheduler, worker, DB, Redis, replier, liker — run in isolated containers.

Docker Compose FastAPI

CLI Reference

Run Zirelia manually with full parameter control

python main.py [OPTIONS]

--platform

Target social platform

twitter instagram (planned) all

--mode

Post type

hybridText + AI-generated image (default)

textText-only post

--topic

Optional topic hint. Omit for fully autonomous mode.

"morning coffee ritual"

--dry-run

Simulate the full pipeline without posting to Twitter/X. Perfect for testing.

Coming Soon

Roadmap features — contributions welcome

🖥️

Control Dashboard

Streamlit/React UI — manage persona, review posts, see analytics

Phase 3

📸

Instagram + LinkedIn

Multi-platform posting with platform-specific content adaptation

Phase 4

🎤

Voice (ElevenLabs)

Audio content — the virtual influencer gets a voice for Reels and TikTok

Phase 5

🎬

Video Generation

Short-form video clips for Reels, TikTok, and YouTube Shorts