I just finished building something I genuinely think is the most complete LinkedIn automation system available as an n8n bundle right now. Not a scraper. Not a connect bot. The full thing. 10 workflows working together as one coordinated agent — it wakes up at 7am, logs into LinkedIn once, stores the session, and from that point your account is running on autopilot for the rest of the day. Here is what it actually does: It finds your 2nd-degree target prospects, visits their profiles the day before connecting so LinkedIn sees human behaviour, then sends a personalised connection request with a custom note. When someone accepts, it detects that automatically and enrolls them into a 4-step DM sequence — opener on day 1, a value message on day 4, a case study angle on day 10, and a soft close with your Calendly link on day 18. Every single message is written in real time by Claude AI based on that person's actual headline and context. Meanwhile it is reading your inbox every few hours, classifying every reply as HOT, WARM, COLD or BOOKING READY, drafting a contextual response and sending it. Hot leads get escalated to a separate booking agent that writes a personalised Calendly DM and sends it automatically. On top of that it publishes LinkedIn content 3 times a week — Claude writes the post in your voice, Playwright posts it without you touching anything. And every Sunday it withdraws pending invites older than 14 days so you never hit LinkedIn's 500-invite cap. The whole thing is coordinated by a master orchestrator that checks your daily quota and ramp schedule every 30 minutes and decides what runs, when, in what order. Nothing fires without permission. Telegram notifies you at every meaningful event. This is not a template. This is a fully wired production system — 10 importable n8n workflow JSONs, a complete Google Sheets CRM schema, Dockerfile for Playwright, and a setup README that walks you through every environment variable and the exact import order. Built on n8n + Playwright + Claude API + Google Sheets. White-label ready — swap the credentials and you have a new instance for a new client.

Here's what it does unattended: WF1 — Scraper & Connect Searches LinkedIn for 2nd-degree targets by keyword, filters out anyone already in the CRM, sends personalised connection requests with a custom note. Runs every weekday at 9am. Max 15 requests/day with human-like delays built in. WF2 — DM Inbox Agent Reads new messages every 2 hours. Claude classifies each reply as HOT / WARM / COLD / BOOKING_READY. Drafts a contextual reply and sends it via Playwright. If the score hits 8+, it automatically triggers WF4. WF3 — Content Publisher Rotates through 8 topic angles, Claude writes the post in the account's voice, Playwright publishes it to LinkedIn. No manual posting. Runs Mon/Wed/Fri at 8am. WF4 — Lead Qualifier & Booking Hot leads get deep-scored against their full CRM profile. Claude writes a personalised DM with a Calendly link. Playwright sends it. You get an email notification. The only thing you touch is the call itself. Stack: n8n (orchestration) · Playwright (browser automation) · Claude API (content + scoring) · Google Sheets (CRM) · Cal.com or Calendly (booking) Happy to share the workflow JSONs if useful.

I know this might sound a bit unusual, but I want to be real. I’ve been building for the past 2 years. I’ve worked with clients, delivering marketing + AI solutions, and I understand how this game works. ( I have a small team) And today I realized something I’m tired of doing this alone. I don’t want to keep building solo. I want someone I can trust to actually scale something big together. I’m looking for someone specific: – obsessed – has experience – has had clients – knows how agencies work I’m 21, so ideally you’re in that 20–30 range. If that’s you, don’t overthink it. Just reach out. Let’s go all in.

I'm 21 and started my own AI automation agency 2 months ago. My first client came through my network, and the next ones came from posting LinkedIn videos about what I was building. But I wanted something that scales without waiting for inbound. So I built a cold email automation pipeline from scratch. It handles the full flow: - Lead discovery using web scraping (Firecrawl + Google Maps data) - Email verification before any email goes out - AI-powered research per lead: website scrape, competitor analysis, needs analysis - Personalized 3-email sequences generated by AI with A/B testing on - subject lines, approaches and styles - Automated follow-up timing (day 1, day 3, day 7) - Reply detection via IMAP with AI classification into 6 categories - Human approval dashboard so nothing sends without review - Lead recycling for non-responders after 60 days The key insight: cold email is not about volume. It is about deep research per lead so the AI has enough context to write emails that actually feel personal. Every lead goes through a 7-step research process before a single word gets written. I put together a complete guide of 26 pages (PDF attached) that covers the full architecture, the actual AI prompts I use, SMTP setup, warmup strategy, deliverability tips, and the 10 mistakes I made building this. Hope it helps someone here!

Anthropic just dropped Computer Use for Claude Code, which lets it natively control your mouse, keyboard, and take screenshots to interact with any app on your desktop. In this video I walk through how to enable it, demo it sending files through ClickUp and controlling my Mac remotely from my phone using Dispatch, and cover the current limitations you should know about. It's in research preview right now but there are already some seriously cool use cases here.