How to reach out and actually get replies

Teach the tool how you talk before you talk to anyone

Start with what happens automatically. When you connect your X/Twitter and LinkedIn accounts, ReacherX reads your own public posts in the background and learns how you write. You'll see a "Learning style" badge in the app header while it works. The goal is simple: when △ Agent drafts a message or a comment for you, it should sound like you wrote it, not like a machine paraphrased you.

Treat that as a base, not a guarantee. The agent can still miss your voice on a given draft, and you're the only one who will notice. When a draft sounds off, don't shrug and edit silently. Tell △ Agent deliberately: "never open with my product", "I write short", "no emojis in the first line". A deliberate instruction is stronger than an inferred style, and it sticks.

Before you contact a single person, save how you like to communicate: the tone you use, the things you never say, the way you open conversations, what has worked for you before. Then test it. Ask for a sample plan for one person and check whether the draft actually sounds like you. If it doesn't, correct it until it does. A draft that sounds like a template gets treated like a template.

One fact worth knowing: what △ Agent learns is scoped per workspace. The style it picks up and the instructions you save belong to the workspace you're in. Tell it once here and it won't automatically carry into your other workspaces. New workspace? Re-teach the important parts, or just tell it again; it takes a minute.

If you're new and don't know your style yet, say so. Ask △ Agent to brainstorm approaches for your situation, and ask it for best practices: it knows the guides on this blog, so it can point you to the right post and summarize what matters, like how voice notes work or how to save instructions it should remember.

Research like you mean it

A message that says "I saw your profile" says nothing. A message that shows you understand someone's work gets read.

Before reaching out, know at least three things: what they've recently posted, what they've built or worked on, and whether you share any context (a community, a mutual connection, a common problem). Read their actual posts, not just their bio.

This is the part ReacherX does best. Every match comes with the posts and background behind it, so you can ground your message in something specific instead of a job title. If you can't find a specific reason to contact someone, that's information too: maybe they're not the right person yet.

Show up in public first

The warmest cold message is one that follows a public interaction. Follow the person. React to their post. Leave a comment that adds something rather than "great post!". Do this a few times over days, not minutes.

A comment that works looks like this:

It references their situation, adds a real experience, and ends on a question that's easy to answer. No pitch anywhere in it.

By the time you land in their DMs, your name is faintly familiar. You're not a stranger asking for something; you're someone who has been useful in a place they already hang out.

△ Agent can help here too: likes, replies, and comments on X/Twitter, reactions and comments on LinkedIn. Keep it genuinely useful, and keep review on for anything that posts publicly under your name.

Then, and only then, talk about your product, service, or ask. If the relationship is real, the pitch is a continuation. If you pitch on day one, you were never building a relationship, you were waiting in line.

Write short, format for skimmers

People read outreach on phones, between tasks, in notifications. Three rules that consistently get replies:

• Keep it under three sentences when you can. One observation, one reason you're writing, one easy question. Clarity beats completeness.
• Format for scanning. Short lines, a line break between ideas, bullets where lists fit. A wall of text answers itself: "later".
• Think about the first line and the last line. Most platforms truncate, so the opening sentence is your headline, and the closing line is what sits in the notification. "Hi 👋 @name" beats "Dear Sir or Madam". A closing question gives them something to answer with one tap.

A full first message, ready to adapt:

Three beats: you read their post, here's what this is, here's an easy question. That's the whole structure.

Fake your formatting with unicode

DMs are plain text. No markdown, no bold button, no code font. The trick people use is unicode characters that look like formatting. Use them for a word or two as an accent, never for whole paragraphs:

Inside a real message it looks like this:

Pick one or two accents per message. A whole message in fake-bold is harder to read, not easier, and some devices render the exotic glyphs unevenly, so stick to the common ones when in doubt.

Show, don't tell

Words are cheap right now, so effort is the differentiator. The single highest-leverage thing you can do is make something for the specific person: a two-minute demo of their exact use case, a screenshot of the thing you'd build for them, a sketch of the workflow you're proposing. Attach it to your request and let △ Agent include it in the plan.

A generic pitch describes. A personal demo proves. It says "I spent an hour on you specifically", and almost nobody does that, which is exactly why it works.

Use your voice when it counts

Text is safe, and safe messages blend together. A voice note changes the channel: your tone, your pauses, your actual personality. It signals that a real human is on the other end, which in a feed full of automation is its own message.

Use it when it adds something: a warm thank-you, a complicated explanation, a first hello after good back-and-forth. Record it yourself in the conversation composer. Thirty honest seconds outperforms a perfect paragraph.

Borrow trust, honestly

People trust their friends' judgment more than your pitch. If someone in their circle uses what you're offering, say so, and make sure it's true. "Ava, who you worked with at Loopstack, mentioned you were looking into this" opens a door that no feature list can.

The same goes for their own content. If a specific post made you reach out, link it. "Your post about losing weekends to invoicing is why I'm writing" shows you did the reading, and it gives them a concrete thread to pull on.

Be a person, not a brand voice

Perfect copy reads like a system wrote it, because usually one did. Small human signals change how a message lands: a typo you didn't correct, "???" when you're genuinely curious, "..." when you're thinking out loud, a ";)" when it fits. Humans are imperfect in exactly these ways.

The same goes for message bubbles. Real people don't send one dense paragraph; they send a short opener, then the substance, then a question. Split yours the same way, and think about the order: platforms preview the first bubble and notifications show the last, so put the hook at the start and the easy question at the end.

Most automation sends one bloated block, and people have learned to smell it.

After you hit send

• Give it time. A reply in three days is normal; an hour is luck.
• If there's no reply, one short, gracious follow-up is fine. Two is pushing it.
• If they say no or go quiet, accept it and keep the door open. People remember how you exit.
• Ask △ Agent to remember what mattered about this person, scoped to them: "Heads-down until their launch, try again in October." It comes back to you in that thread, next time you open the conversation.

A follow-up that doesn't nag:

One line, an easy exit, and a real compliment. That's the whole follow-up.

The unfair advantage

None of these practices are secrets. The difference is that doing all of them, every time, for every person, is expensive. That's the part a tool should absorb: the research, the memory of your style, the media pipeline, the review before anything sends. What's left for you is the part only a human can do: showing up with context, care, and a reason.

Do that consistently and your reply rate stops being luck. It becomes a practice.