Agent Maya

Everything you need to know about Maya. This article walks you through exactly how she works so you can see what’s happening behind the scenes and how to get the most out of the process.

The challenge getting sales calls

Sales outreach in 2025 isn’t as simple as sending a bunch of cold messaging and praying someone replies and books a call. People's behaviours have changed, a lot. And while there are endless tools and AI platforms, many teams still get stuck on exactly knowing what to say, when to say it, and how to keep track of every opportunity. And no one has fully delegated this to AI successfully, yet.

That’s why we've built Agent Maya. She’s an AI agent designed to run outreach on LinkedIn in a way that feels natural for the prospect but still drives results for you and your team. Instead of spamming generic compliments or rushing straight to a pitch, Maya follows a tested strategy that warms people up gradually and then helps you guide conversations toward a call.

How Maya starts outreach

Maya's auto-pilot mode works from the lists you provide. We'll provide her a list of your ideal buyers from

LinkedIn Sales Navigator
a CSV file
or a custom search on LinkedIn

From there, she starts moving people from cold prospects to warm prospects using a deliberate step-by-step sequence.

This process always begins with light engagement:

Maya views the prospect’s profile.
She likes their most recent post.
Then she waits two days.
After those two days, she repeats the same sequence , another profile view, another post like.

This creates small moments of familiarity so that when she eventually sends a connection request after 2-3 days, the prospect is more likely to recognise your name, and therefore accept.

When Maya sends the connection request, she does it without adding a note. The reason is simple. All the data shows that people are more likely to accept a request from someone they’ve seen pop up a couple of times than from a stranger sending a cold pitch. Once that connection is accepted, Maya waits another two days before sending the first direct message.

What Maya sends in her first message

Once Maya has warmed up a lead, she sends something like this:

Hi John. [personalised icebreaker] :)

[Qualifying question]?

Detail on this are in our outreach playbook, linked here - which I'd strongly recommend you read if you haven't already.

That’s what Maya's auto-pilot mode will do once you’ve already proven your messaging and knosw exactly what resonates with your ideal customer profile.

In the early days though it’s a different story.

You don’t want to burn through a valuable list of prospects by sending untested messages that flop.

Instead Maya opens with something neutral like Hey John. Great to connect with you.

This simple first touch marks the connection in your inbox and gives you time to experiment with different icebreakers and qualifying questions using Maya's co-pilot. Once you see which approaches get the right responses, we feed those insights back into Maya, so she can handle them on autopilot without risking the quality of your outreach.

Why the early stage stays hands-on

In the early days of any campaign, it’s better to stay proactive instead of automating everything. At this stage, you’re still learning about your ideal customer profile and testing different approaches to see what sparks replies.

Sending out mass messages or blanket compliments too early can backfire. Without understanding the prospect’s goals and pain points, even small missteps can work against you. The Flow team will provide you plenty of data to help optimize your messaging before scaling. Once that foundation is in place, Maya can take care of it.

How Maya keeps outreach safe

LinkedIn has strict rules around connection limits and spam detection.

To keep everything safe, Maya will send a maximum of 15 connections requests a day. That gives you a cushion of 5 additional connection requests to play with when you're active on LinkedIn. Maya will also automatically withdraw any connection request that hasn’t been accepted within 14 days. This does two things.

First, it avoids LinkedIn flagging your account for using automation.
Second, it gives you another chance with that same prospect at a later date. LinkedIn allows you to resend a connection request after three weeks.

Maya also uses AI to clean details like first names and company names before messages are sent. This removes awkward errors and keeps your outreach feeling human, and professional. Typically, she'll run her tasks at activity at 15-minute intervals to mirror human-like behaviour.

Understanding Auto-pilot Mode

Autopilot mode is what handles the warm-up sequence only. It’s responsible for the quiet touches that make you familiar before the first message lands.

viewing a profile
liking posts
sending the connection request
and finally dropping that simple first DM once the connection is accepted.

She'll also continue to like posts and view their profile after the first message is sent, to help prompt replies.

This mode is perfect for running in the background while you focus on other things. It keeps outreach consistent, follows LinkedIn’s best practices, and ensures you’re not manually spending hours adding your ICP to your network, and on other basic engagement steps. But it stops there. Once a reply comes in, that’s when copilot takes over.

Understanding Co-pilot Mode

Copilot mode is where Maya becomes more collaborative.

Instead of automating replies, she works alongside you to generate them. When someone replies to your first message, copilot suggests thoughtful responses based on the outreach playbook, your profile, and information Maya's reading on the prospect’s profile. She'll generate the reply. Then you can edit or tweak these before sending. This still give you full control, without the headache of working out what to say.

If you're beginning your outreach journey, and sending simple messages like:

‍Hey John. Good to connect.

This gets your ideal prospects into your inbox so you can treat them like a checklist and work through each one personally. Maya's copilot can then be used to create personalised icebreakers and qualifying questions to get the conversation going.

I've mentioned it before, but I'll say it again. This is especially helpful early on when you’re still testing which approaches resonate most with your ideal customer profile. The feedback loop here is powerful, you experiment with different angles, see what gets traction, and then those proven messages eventually feed back into your process of generating replies, and guiding conversations to a call.

Staying organised

The hardest part of LinkedIn prospecting isn’t just sending messages, it’s keeping track of every conversation once people start replying. Flow’s CRM keeps it simple. It shows you which prospects are active, nudges you when follow-ups are due, and makes sure no opportunity gets buried.

With both modes working together and a clear view of your pipeline, you can steadily move prospects to booked calls without losing momentum.

Bringing it all together

So, quick recap - Maya’s outreach begins with light touches that build familiarity, moves into a simple first message, and then shifts into copilot. This is where you begin testing different icebreakers and qualifying questions, learning which snippets guide conversations toward a booked call.

After a couple of weeks you’ll find a rhythm.

Certain replies will repeat themselves.
Certain approaches will consistently move people closer to a call.

Those proven snippets become part of your personal playbook, ready to reuse with Maya's co-pilot across future prospects - so outreach gets smoother and faster over time.

Our internal KPI is to help you reach two to three calls a week. To get you there we provide hands-on support, helping you to optimize your profile, guiding your content, and refining your outreach so every reply feels natural and generates you pipeline.