toflow.ai Logo
LinkedIn Use Cases

Connect and Send Follow-up Messages

Claude writes personalized connection requests, follows up once accepted, and logs every interaction automatically.

Connect and Send Follow-up Messages

A LinkedIn connection request without a note gets treated like a cold email with no subject line. Adding something relevant dramatically improves acceptance rates, but writing a personalized note for every prospect is not sustainable at volume. And even when requests are accepted, most outreach stalls because the follow-up does not happen fast enough.

This workflow handles both. Claude builds a two-step sequence in toflow: a connection request with a personalized note, then a follow-up that fires only after the request is accepted. You describe who you are targeting and why. Claude sets up the sequence, writes the content for each person, and toflow runs it.


How to connect toflow with Claude

toflow.ai has a built-in MCP server that lets Claude control your LinkedIn outreach directly. It works with Claude Desktop, Cursor, Windsurf, or any MCP-compatible client.

Step 1: Connect your LinkedIn account to toflow

Log into the toflow app and go to Settings, then Accounts, then Connect LinkedIn. This is the account Claude will use to send requests and messages on your behalf.

Step 2: Connect toflow to Claude Desktop

  1. Open Claude Desktop, go to Settings, then Connectors, then Add Custom Connector
  2. Set the Name to toflow.ai and the URL to https://mcp.toflow.ai/mcp, then save
  3. Click Connect. Claude will open a browser window to authorize your toflow account
  4. Once authorized, start a conversation. Claude can now build sequences, send connection requests, and log every interaction in toflow

Using Cursor, Windsurf, or another MCP client? Add this to your config file instead:

{ "mcpServers": { "toflow": { "type": "http", "url": "https://mcp.toflow.ai/mcp" } } }

Once connected, describe what you want in plain language and Claude handles the rest.


What you give / What you get

What you give

  • A list of LinkedIn profiles (from a search, CSV, or toflow)
  • Context about who you are reaching out to and why
  • Your preferred tone and daily sending limits

What you get

  • A sequence in toflow with a connection request step and a follow-up step
  • A personalized connection note and follow-up written per person by Claude
  • Follow-up sent only after the connection is accepted, not to everyone
  • Every step logged in toflow automatically

What this workflow does automatically

Claude creates a two-step sequence in toflow. The first step is a LinkedIn connection request with a personalized note. The second is a follow-up message, with a gap between the two. toflow monitors the connection status after each request, and the follow-up only triggers once the request is accepted. People who ignored it never receive the second message.

Before enrolling each prospect, Claude reads their profile. It looks for something specific to reference, a recent post, a company update, or something in their role description. The connection note and follow-up are written fresh for that person, not pulled from a template.

Once the sequence is running, toflow spaces requests across the day within your daily limit. Every send, acceptance, and follow-up logs automatically. Replies land in your toflow inbox, categorized and ready for you to respond.


How teams use this workflow

SDRs building pipeline from LinkedIn searches

Run a LinkedIn search for your target persona, export the results, and give the list to Claude. Claude writes a different connection note for every person based on their profile. You review the batch, approve it, and the sequence runs. When someone accepts, the follow-up fires automatically. You only step in when there is a real reply.

Founders doing outreach without a team

Founders using this workflow run consistent LinkedIn outreach without spending hours on it each week. Claude handles the research and writing. You stay in control by reviewing before anything sends. The follow-up happens even when you are focused on something else.

Agencies running campaigns across multiple clients

Each message reflects the specific prospect and the client's context, not a shared template. Volume stays consistent across large lists without quality dropping.


Step by step

  1. Add a prospect list to toflow, from a LinkedIn search, CSV upload, or existing contacts
  2. Tell Claude who you are targeting and why you are reaching out
  3. Claude builds a two-step sequence in toflow: connection request, then a follow-up with a gap
  4. Claude reads each profile and writes personalized content for both steps at enrollment
  5. Review the sequence and content before anything sends
  6. toflow sends connection requests at a human pace within your daily limits
  7. When a request is accepted, the follow-up fires automatically
  8. Replies land in your toflow inbox, ready to respond

Best practices

Start with 15 to 20 connection requests per day and increase gradually. LinkedIn monitors sudden spikes in activity and conservative limits protect your account.

Give Claude specific context about your offer and why you are reaching out to this persona. Vague input produces generic notes. The more specific you are, the more the message sounds like it came from you.

Review the notes before the sequence runs. Claude writes from profile data, but a quick scan before approving catches anything that needs adjusting.

Do not shorten the gap between the connection request and the follow-up. A message that arrives the moment someone accepts reads as automated. The timing is part of what makes it land well.

Handle replies from the toflow inbox, not LinkedIn directly. Everything stays in one place and the full context is there when you respond.


Ready to try it?

Connect toflow to Claude, describe who you want to reach, and have your first sequence running in minutes. No coding required.

Book a demo. 2 weeks free, no credit card required.


FAQ

Frequently asked questions

Run "Connect and Send Follow-up Messages" with toflow.ai