Type
TypePersonal project
Portfolio
AI SYSTEMS & WORKFLOWS
Human-guided AI LinkedIn publishing
Built an AI-assisted workflow that turns weekly work into high-quality LinkedIn content at scale, without burning time or sacrificing quality.
Last updated: 2026-05-17
3x posts/wkPublishing cadence
~2+ hrs/wkSaved on content creation
Project snapshot
Problem
ProblemInconsistent posting and too much time spent turning ideas into quality drafts.
Solution
SolutionAI-assisted system with clear stages, gates, and review to produce better content, faster and consistently.
Outcome
Outcome3x posts/week | 2+ hours/week saved
Tools & Systems
Tools & SystemsChatGPT, Codex, OpenAI API, Postgres, Docker, Next.js
The problem
Great ideas, poor system.
- Ideas lived in scattered notes and documents
- Turning thoughts into posts took too much time
- Inconsistent quality and posting cadence
- No reusable system, starting from scratch every week
The approach
Design a system, then scale it.
- Capture weekly accomplishments and raw notes
- Extract the strongest story with AI
- Draft from a narrative direction
- Review and refine for quality and clarity
- Publish and learn from engagement
The system
A repeatable workflow from raw notes to published posts.
Capture
Extract
Draft
Review
Publish
ChatGPTCodexOpenAI APINext.js + PostgresWeekly Review CadenceLinkedIn Analytics Feedback
Key insights
- Start with story signal, not blank-page prompts
- A clear narrative direction makes drafts better and faster
- Human review is the quality multiplier
- Systems > motivation for long-term consistency
Lessons learned
- Invest in orchestration earlier to reduce manual steps
- Build review gates smarter for higher signal feedback
- Design for iteration, the system keeps getting better