🪶 The Aurelian Content Calendar

AI-Powered Instagram Content Engine for Luxury Brands

A multi-agent AI system built with CrewAI that autonomously researches Indian cultural moments, analyses luxury fashion trends, and generates a complete 30-day Instagram content calendar — captions, hashtags, and post strategy included.

---

What It Does

Given a target month, four specialised AI agents collaborate in sequence to produce a fully formatted content calendar tailored to an Indian luxury fashion brand:

1. Cultural Researcher — Scours the web for Indian festivals, public holidays, wedding seasons, and lifestyle moments relevant to an affluent audience
2. Trend Analyst — Researches luxury fashion trends, competitor activity, and trending hashtags in the Indian market
3. Content Strategist — Maps cultural moments and trends to a strategic 30-day posting plan, balancing Reels, Carousels, Static Images, and Stories
4. Caption Writer — Writes final captions and curated hashtag sets in the brand's voice: old money, understated, rooted in Indian heritage

The output is a structured markdown file with a post plan for every day of the month.

---

Architecture

┌─────────────────────┐     ┌──────────────────────┐
│  Cultural Researcher│────▶│   Trend Analyst       │
│  + SerperDev Search │     │   + SerperDev Search  │
└─────────────────────┘     └──────────────────────┘
           │                          │
           └──────────┬───────────────┘
                      ▼
           ┌──────────────────────┐
           │  Content Strategist  │
           │  (30-day plan)       │
           └──────────────────────┘
                      │
                      ▼
           ┌──────────────────────┐
           │   Caption Writer     │
           │   (Pydantic output)  │
           └──────────────────────┘
                      │
                      ▼
           output/content_calendar.md

Process: Sequential
LLM: OpenRouter (GPT-4o-mini / configurable per agent)
Search: SerperDev API
Output schema: Pydantic-validated ContentCalendar model

---

Sample Output

# The Aurelian — Content Calendar: May 2026

---

## Day 1 — May 1, 2026

**Post Type:** Carousel

**Theme:** Buddha Purnima: Celebration & Reflection

**Caption:**
In stillness, we find enlightenment. As we honour Buddha Purnima, let us
reflect upon the virtues of peace and spirituality amidst the tapestry of
luxury. ✨

**Hashtags:** #TheAurelian #BuddhaPurnima #LuxuryLifestyle #SpiritualElegance
#HeritageFashion #LuxuryIndia #TimelessDesign #CulturalWealth

**Why this day:** Celebrating the enlightened values of Buddha while showcasing
a luxurious collection for spiritual events.

---

Tech Stack

| Component | Technology | |---|---| | Agent Framework | CrewAI | | LLM Provider | OpenRouter | | Web Search | SerperDev | | Output Validation | Pydantic v2 | | Package Manager | uv | | Language | Python 3.12 |

---

Project Structure

aureliancalender/
├── src/
│   └── aureliancalender/
│       ├── crew.py          # Agent and task definitions
│       ├── main.py          # Entry point
│       └── config/
│           ├── agents.yaml  # Agent roles, goals, backstories, LLM config
│           └── tasks.yaml   # Task descriptions, context chaining
├── output/
│   └── content_calendar.md  # Generated calendar (git-ignored)
├── .env                     # API keys (git-ignored)
├── .gitignore
└── pyproject.toml

---

Getting Started

Prerequisites

• Python 3.12+
• uv installed
• An OpenRouter API key
• A SerperDev API key

Installation

# Clone the repo
git clone https://github.com/yourusername/aureliancalender.git
cd aureliancalender

# Install dependencies
uv sync

Configuration

Create a .env file in the project root:

OPENROUTER_API_KEY=your-openrouter-key-here
SERPER_API_KEY=your-serper-key-here

Run

crewai run

The calendar will be generated at output/content_calendar.md.

To change the target month, edit main.py:

inputs = {"month": "June 2026"}

---

Key Design Decisions

Why sequential process? Each agent's output is the next agent's context. The caption writer needs the cultural research, trend data, and strategic plan all in scope to write coherent, well-timed captions.

Why Pydantic output? Structured validation ensures all 30 days are present and correctly formatted before writing to file. If the model truncates or hallucinates, the schema catches it.

Why OpenRouter? Flexibility to swap models per agent without changing code — upgrade the caption writer to a higher-output model while keeping cheaper models for research tasks.

---

Roadmap

• [ ] Add image prompt generation per post (DALL-E / Midjourney prompts)
• [ ] Export to Notion or Google Sheets
• [ ] Telegram/Slack delivery of the finished calendar
• [ ] Support multiple brand voices via config
• [ ] Week-by-week chunking to improve 30-post reliability

---

License

MIT