Welcome to session 47! This week Yashveer Yadav shares how he built a practical LinkedIn automation workflow for maintaining professional relationships without turning LinkedIn into a spam machine.
Yashveer is an Associate Partner at Antler India. He works at the intersection of capital, startups, and private-market opportunities, and has been experimenting with AI-powered personal automation for relationship management.
Session Overview
Yashveer walks through the system he built to automate two daily LinkedIn workflows: reviewing incoming connection invites and sending birthday wishes. The core idea is simple: let automation handle the repetitive browsing, summarization, and clicking, but keep a human approval step before anything is accepted or sent.
Key Takeaways
- Relationship automation should stay human-in-the-loop: The workflow summarizes invites and birthday messages, but Yashveer still chooses what gets accepted or sent.
- Browser automation is useful when APIs are unavailable: Playwright controls a real Chrome session, so the workflow behaves like normal browsing instead of relying on restricted LinkedIn APIs.
- Local-first design keeps the blast radius small: Session cookies, logs, skip lists, and sent-message history stay on the user's machine.
- Small scripts are easier to debug: Invite automation and birthday automation are separate scripts, which makes LinkedIn UI breakages easier to isolate.
- AI coding tools can help non-engineers build real workflows: Yashveer used Claude to build, debug, and iterate on a system that works for his daily routine.
Topics Covered
Why Automate LinkedIn
Yashveer explains why birthdays and connection invites are useful relationship moments. The goal is not to automate generic outreach; it is to make it easier to keep warm professional connections active.
Daily Workflow
The system runs every morning, finds pending invites and birthdays, presents a selectable summary, and proceeds only with the people Yashveer approves. It can also handle belated birthday wishes and avoid duplicate messages.
Technical Architecture
The workflow uses Playwright with Node.js, Chrome session cookies, macOS launchd scheduling, local JSON logs, skip lists, and deduplication state. The scripts are kept separate so invite automation and birthday automation can evolve independently.
Live Demo
Yashveer demonstrates invite acceptance and birthday-message automation live, including the prompt UI, LinkedIn navigation, and how the script handles UI issues when LinkedIn changes.
Q&A Highlights
- LinkedIn safety and why browser automation behaves differently from API calls
- Local privacy, session cookies, and machine-local logs
- Moving similar workflows to the cloud while preserving approval steps
- Improving birthday-message personalization without making it too casual
- Extending the approach to jobs, content, and broader relationship-building workflows
Here's the entire recording of the session.