< All Blog Posts
Product Published March 1, 2026

AI LinkedIn messages: Co-pilot vs Autopilot for sales

I hear “AI outreach” used for two very different jobs. Here is how Autopilot and Co-pilot map to LinkedIn sales in Flow AI, and how I use them without handing the whole relationship to a bot.

Darren Alderman

Darren Alderman

Co-founder, Flow AI

Professional workspace with laptop
  • Autopilot as lists: warming, connection requests, and backend campaign steps
  • Co-pilot in Messages: AI drafts from profile and thread, reply playbook, you hit send
  • How I pair them so scale does not replace judgment in the inbox

In Flow AI, a list is your Auto-pilot campaign: it runs the structured LinkedIn steps (profile visits, likes, connection requests) on a schedule the system enforces. Co-pilot lives in Messages. That is where AI drafts replies from the prospect’s profile, the full thread, and your offer context, then you send. Same product, two different jobs, and I treat them differently on purpose.

What Auto-pilot covers on LinkedIn

When I say Auto-pilot here, I mean the list-backed campaign, not “AI wrote this DM.” You build a list, assign sender accounts, add people from Search or CSV, and turn the list on. Behind the scenes we run a fixed multi-step sequence: visits, post likes, waits, then connection requests, with daily caps (for example 15 connection requests per account per day once warmed) and a 9am to 6pm local window so behaviour still looks human.

That is the right place to automate repetition I would not want a rep to do by hand hundreds of times. It is not where I want AI guessing the final wording of a sensitive first touch to an executive. Auto-pilot gets people into my network and keeps early steps moving; it does not replace how I think about a live conversation.

What Co-pilot does when someone replies

Open Messages from Connections or the Messages tab and you get the two-pane layout we designed on purpose: conversation on the left, enriched profile on the right, sender account visible so the team does not cross wires. In the composer I can type manually, drop snippets, or use custom prompts. When I ask for an AI draft, it reads that profile plus the whole conversation history, pulls in the offer context from settings, and follows our reply playbook so the suggestion is not a generic paragraph.

Critical detail: the user reviews before sending. There is no auto-send from Co-pilot. That is the product decision I stand behind for LinkedIn sales. You get speed and structure; you keep accountability. If you want reusable angles and phrasing to tighten those drafts further, our piece on AI reply templates that actually convert to meetings fits this workflow well.

Auto-pilot fills the top of the funnel. Co-pilot helps me answer like a human when people write back.

How we separate campaign automation from inbox drafting

When I run both in the same pipeline

My typical week: Auto-pilot lists keep new qualified prospects moving through warming and connection requests while I live in Messages for anyone who replies or accepts. Dashboard metrics (acceptance rate, reply rate, active conversations) tell me whether the list side is healthy; the inbox tells me whether my replies turn into next steps.

Settings includes a Co-pilot tab for custom prompts and dynamic snippets. That is how teams keep “our point of view” consistent without making every message sound copied and pasted. Scheduling a follow-up DM for later still goes through the same compose path, so Co-pilot and timing work together.

Mode Best for Who owns the last mile
Auto-pilot (lists) Repeatable early-stage steps across many prospects System sequence within LinkedIn limits
Co-pilot (Messages) Replies, follow-ups, nuanced asks after a human has engaged You, after reviewing the draft

How I choose speed versus control

I reach for Auto-pilot when I have a clear slice of ideal buyers and I need the same steps done across hundreds of profiles without babysitting every click. I lean on Co-pilot when tone, objection handling, or a specific detail in the thread matters more than raw speed.

Neither mode is “set and forget.” Auto-pilot still needs sensible lists, sender rotation, and respect for warm-up and daily caps. Co-pilot still needs a quick human edit so the send sounds like your team, not a model. Used together, I get scale where it is safe and judgment where it earns revenue.

Next step: try Flow AI

If this split makes sense for how you sell, open the app and run a list on Auto-pilot, then open a thread and generate a draft with Co-pilot. Brush up phrasing with AI reply templates that actually convert to meetings, then try Flow AI free and see both modes on your own accounts.

Frequently asked questions