What the LinkedIn Message Writer does
Agent Maya writes LinkedIn connection notes and DMs in your voice and sends them for you. Every message is drafted from the prospect's real context — their role, their company, something from their recent activity — not swapped into a template. You can work from Flow's Co-pilot, or ask Claude or Cursor to write and send a note through Flow's LinkedIn MCP.
Who this is for
- SDRs and AEs who live on LinkedIn but can't write genuinely personalised messages at the volume they need.
- Founders who want every note to sound like them, without spending an hour a day in the message box.
- Teams already working in Claude or Cursor who'd rather draft and send LinkedIn messages from their AI assistant.
What Agent Maya takes off your plate
- Connection request notes written from each prospect's role, company, and LinkedIn activity.
- DMs and replies drafted in your voice using the full conversation history.
- Saved prompts and snippets so a warm reply, a post compliment, or a follow-up is one click away.
- Sending on your behalf — from Flow's Co-pilot or from Claude and Cursor via the LinkedIn MCP.
- Every message logged against the lead so you always know what went out.
How Agent Maya does it
- It reads the prospect's context: Maya pulls the prospect's LinkedIn profile, your message history, and your offer context before writing a word.
- It drafts in your voice: Applying your reply playbook and any prompt you pick, Maya writes a suggestion straight into the message box — click Draft again for a fresh take, or edit it directly.
- You review, then send: Nothing goes out until you approve it. Send now, or use the arrow next to Send to schedule the message for later.
- Or drive it from Claude or Cursor: With Flow's LinkedIn MCP connected, ask Claude to write and send a personalised note to this prospect and Maya does it — reading the profile and sending the message through your Flow account, with your confirmation.
- Everything is tracked: Each note and reply is recorded against the lead so your pipeline stays up to date.
What to expect
Every prospect gets a message that reads like you sat down and wrote it — because the words come from their real context, not a mail-merge field. Reps stop staring at a blank message box and start sending.
Because you can draft and send from wherever you work — Flow's Co-pilot or Claude and Cursor via the MCP — personalised LinkedIn messaging stops being the bottleneck and becomes something you actually keep up with.
How Flow AI sets this up
- Connect your LinkedIn account and set your voice, reply playbook, prompts, and snippets in Flow.
- Create a Flow API key and add Flow AI as a custom connector in Claude, Cursor, or any MCP client.
- Start prompting — ask your AI assistant to write and send a personalised note, and review before it goes out.
FAQ
Does Maya really write in my voice?
Yes. Every draft is generated on demand from the prospect's profile, your full conversation history, and your offer context, then written using your reply playbook so it stays in your voice — not a template with the name swapped in.
Can I write and send from Claude or Cursor?
Yes. Flow AI runs an MCP server, so once you add it as a connector you can ask Claude or Cursor to write and send a personalised message to a prospect and Maya does it through your Flow account. Anything outbound asks for your confirmation first.
Does anything send without me approving it?
No. Flow is built around Co-pilot mode: Maya suggests the message and you stay in control. In the app nothing sends until you hit Send, and over the MCP outbound actions ask you to confirm first.