🚀 I Built an AI Agent That Writes Viral LinkedIn Posts in My Voice
Most AI writing tools sound the same.
Same hooks. Same tone. Same “AI feel.”
So I built LinkedIn Post Pilot — an AI system that:
- Writes LinkedIn posts in my voice
- Predicts what will perform well
- Learns from real engagement
This isn’t a prompt wrapper.
It’s a self-improving AI pipeline.
💡 The Idea
I wanted a system that:
- Generates high-quality content daily
- Feels personal (not generic AI)
- Improves over time using feedback
⚙️ How It Works
ONBOARD → GENERATE → SELECT → PUBLISH → LEARN
1. Persona DNA (Voice Matching)
- Analyzes my past posts
- Extracts tone, structure, vocabulary
- Injects rules into every generation
👉 Result: content sounds like me, not AI :contentReference[oaicite:0]{index=0}
2. Trend Radar
- Tracks topic momentum (velocity + growth)
- Identifies emerging trends early
3. Multi-Agent System
- Researcher → finds ideas
- Writer → drafts posts
- Critic → scores quality
- Revision → improves output
4. Virality Scoring
Each post gets:
- 3 hook variants
- A score (0–100%) based on:
- Hook strength
- Format
- Engagement potential
- Trend timing
5. Daily Output
Every morning:
- 7–10 ready-to-post ideas
- Hook variations
- Virality scores
Pick one → post in 2 minutes.
6. Post Autopsy (Learning Loop)
After publishing:
- Track likes, comments, shares
- Compare predicted vs actual performance
- Update:
- Scoring weights
- Persona preferences
- Topic memory
👉 The system improves every week :contentReference[oaicite:1]{index=1}
🧑💻 Tech Stack
- Python
- OpenAI / Anthropic / Ollama
- YAML configs
- Pytest
- SMTP / Resend
🧠 Why It’s Different
- Uses your voice, not generic AI
- Learns from real results
- Avoids repeating content
- Uses multi-agent reasoning
🚀 Try It
👉 https://github.com/Sohamp2809/viral-ink
🧠 Final Thought
AI won’t replace creators.
But creators who use AI systems like this will win.
Top comments (0)