Design with May — I Built a $10K LinkedIn Outreach Automation (Claude Cowork)

Source: YouTube Channel: Design with May Duration: 6:52 Views: 51961 · Likes: 1202 Video: Watch on YouTube

In this video, I show you how we turned a simple spreadsheet of podcast hosts into a LinkedIn outreach system that runs in the background.

Using Claude Cowork + ChatGPT, we built an automation that: • Generates personalized LinkedIn connection notes • Pulls the right data from our database • Opens profiles in the browser • Sends requests step-by-step • And brings real conversations back

This isn’t theory. This is a real workflow we built for a real business (PodStar) using data we already had.

If you already have: • A spreadsheet of leads • Past outreach • Partnerships • Warm contacts

You’re probably sitting on the same opportunity.

⏱ Chapters: 00:00 The $10K LinkedIn Automation 00:26 What is Claude Cowork? 01:16 Context – PodStar 01:46 The database strategy 02:28 Building the outreach message 03:38 Build the automation with Claude Cowork 04:12 Testing + running 10 connection requests 05:07 LinkedIn limits you need to know 05:49 Recap

⚠️ Important: If you’re using LinkedIn free, you can only add personalized notes to 5 connection requests per month. Premium allows unlimited personalized notes, but pace your automation responsibly.

—

Key Insights

I built a $10,000 LinkedIn outreach automation using Claude CoWork. And if you already have a spreadsheet of leads, partnerships, or past conversations, you’re probably sitting on the exact same opportunity. [music] In this video, I’m going to show you how we built an outreach system that runs in the background, sends personalized LinkedIn connection notes, and already started bringing conversations back. To build all of this, I used Claude CoWork. It’s an AI desktop agent from Anthropic, which means it can actually do work on your computer, not just chat. It can run multiple tasks at the same time and automate actions in your browser. That’s huge because LinkedIn lives in the browser, so Claude can literally click through the steps for you. So, while you’re living your life, it can read and update spreadsheets, create files, organize folders, and execute browser steps to get the job done. It can also connect to other apps via MCP, which is Model Context Protocol, and it’s basically a universal remote that lets Claude talk to different tools. And this is exactly what we’re doing today. We’re taking a spreadsheet and using it for personalized LinkedIn outreach at scale. Now, let me show you the exact workflow we built. So, here is a bit of context. My husband runs a business called PodStar, where he connects company founders with global podcasts. Over time, he built a pretty large database of podcast hosts, people he either cold pitched through email or collaborated with through his clients. LinkedIn is his main relationship engine. That’s where he strengthens connections and open doors for his clients. So, he already has a spreadsheet full of warm contacts he can use for better outreach. Let me quickly show you what we’re working with. Inside the database we have the podcast host name, the podcast LinkedIn profile link, the podcast name, the podcast theme, meaning the niche, the matched speakers, which client he previously pitched to that show, and the last contact date, which was usually done by email. This is important because when you plan your outreach, you need to look at the data you already have and decide what fields actually matters for personalization. What should the AI reference based on what makes someone reply? The first thing we did was take the data we already had and start planning the prompt we would send to Claude using ChatGPT. We started with the outreach message itself. We planned to send this as a LinkedIn connection request with a note, and LinkedIn has a hard limit of 300 characters for that note. So, we asked ChatGPT to write a message template based on the exact fields we wanted it to pull from our spreadsheet. Here’s the version we chose. Once the message was ready, we gave ChatGPT the workflow step-by-step in plain exact instructions. Here’s the action order. One, generate the message for each row using the template and fill in the missing variables from the spreadsheet. Two, send me the updated spreadsheet and ask for approval before starting outreach. Three, after approval, open each person’s LinkedIn profile using the link in the spreadsheet. Four, click more, connect, add a note, and send the personalized note for that specific row. If we’re already connected on LinkedIn, go back in the spreadsheet, mark it as already connected, and don’t send anything. Then, we took that full prompt and moved into Claude to actually build the automation. And from here, it’s shockingly simple. First, make sure you have the Claude desktop app installed. Open it, and at the top, switch to co-work. Quick side note, to use co-work, you will need the pro plan. So, if you don’t have it yet, set that up before you start. Once you’re ready, just paste the prompt you built in ChatGPT, attach your spreadsheet, and send it. Claude sent the spreadsheet back with the outreach messages it generated using the template and the data from our table. After a quick review to make sure everything looked right, we asked it to test j

Transcript continues…