# Midjourney Acquires Co-Star: AI Art Meets Astrology

> Source: <https://promptcube3.com/en/threads/2878/>
> Published: 2026-07-24 19:43:28+00:00

# Midjourney Acquires Co-Star: AI Art Meets Astrology

Co-Star isn't your typical "newspaper horoscope" app; it leverages NASA data and algorithmic processing to provide hyper-personalized daily insights and compatibility checks. The acquisition, which reportedly closed earlier this spring, suggests that Midjourney is looking to experiment with how generative AI can interpret complex datasets to provide personalized human experiences.

## Why this matters for AI Workflows

From a technical perspective, this is an interesting move in the realm of LLM agent development. Co-Star's core value is its ability to synthesize astronomical data into a specific "voice" or personality—often described as blunt or poetic. Midjourney is likely interested in this specific type of personalized synthesis.

If we look at this as a case study in AI workflow integration, the synergy lies in "multimodal personalization." Imagine an AI agent that doesn't just tell you your horoscope in text, but generates a personalized visual representation of your daily cosmic alignment using Midjourney's diffusion models.

## Technical Implementation: Simulating the "Astrology Agent"

If you are trying to build a similar personalized agent from scratch using current LLMs, the key isn't just the prompt, but the data pipeline. You cannot rely on the LLM's internal knowledge of planetary positions because they hallucinate dates and degrees.

To build a real-world astrology-style agent, your deployment architecture should look like this:

1. **Data Fetching:** Use an API (like a Swiss Ephemeris wrapper) to get the exact planetary coordinates for the user's birth time and current time.

2. **Context Injection:** Pass these coordinates as raw data into the prompt.

3. **Persona Layering:** Use prompt engineering to define the "voice" (e.g., "The Stoic Astrologer").

Here is a basic example of how you would structure the system prompt for a personalized insight agent to avoid generic outputs:

```
{
  "system_prompt": "You are a precise astronomical analyst. You will be provided with current planetary transits (longitude/house). Your task is to synthesize this data into a concise, slightly cryptic daily insight. Avoid phrases like 'The stars suggest' or 'You might feel'. Instead, lead with the planetary tension. Format: [Planet] in [House] -> [Direct Insight].",
  "user_data_context": {
    "sun_sign": "Scorpio",
    "current_transit": "Mars square Pluto",
    "house_position": "8th House"
  }
}
```

## Is this worth following?

For most developers, this acquisition is more of a signal than a tool. It shows that the "AI Art" companies are evolving into "AI Experience" companies. The real value here is the potential for a deeper integration of personalized data and generative media.

If you're building an LLM agent, the lesson from Co-Star is clear: the most successful "personalized" AI apps aren't just wrappers; they are systems that combine proprietary data (like NASA's astronomical catalogs) with a very strong, opinionated brand voice. Midjourney is now positioned to combine that voice with the most powerful image generation engine in the world.

[Next Jensen Huang's Move to Open Models →](/en/threads/2876/)

## All Replies （3）

[@PatFounder](/en/users/PatFounder/)that'd be wild. Imagine prompting "Leo energy" and getting specific color palettes based on the stars lol