The science

The Signl Intent Engine

A proprietary, neural intent engine that reads what your market says in public and decides, person by person, who is ready to talk to you, why, and what to say first.

The formula

One score for who to talk to first

Every person the engine finds gets a Signal score. It is built from four things, and the order in your inbox follows it.

Intent

What they said, as one of nine kinds of intent. It counts for more than fit, because what someone says this week beats who they are on paper.

Fit

How well the person and their company match who you sell to: role, seniority, industry, size and where you sell.

Confidence

How sure the model is of its own read. An unsure read is scaled down, so a guess never outranks a clear signal.

Learning

What your own outcomes taught it. People like the ones who replied to you move up; the kind you keep setting aside move down.

Inside the engine

Five stages between a comment and a conversation

Cheapest first. Each stage only sees the people the one before it kept.

  1. Listen 01

    Every source, every hour

    Signl watches the profiles, posts and searches where your buyers talk, on LinkedIn and X, and collects every comment, reply and reaction as it lands.

  2. Filter 02

    Noise out, before anything is spent

    Rules strip the empty comments, the congratulations, the job hunters and the author talking to themselves. Nobody pays for a lookup on a 'congrats!'.

  3. Classify 03

    A neural read of every comment

    A neural language model reads what each person said against who you sell to, sorts it into one of nine kinds of intent, and states how sure it is. A low-confidence read is set aside, with the reason.

  4. Qualify 04

    The person, the company, the offer

    A verified work email, the employer confirmed from what the profile actually says, and which of your offers fits what they raised.

  5. Write 05

    Five drafts, one judge

    Five first messages on five different approaches. Code throws out anything generic, salesy or invented, and a second model judges the rest. You get the best, and two alternatives.

The nine intents

Not all engagement is equal

A like and a sentence that says "we are replacing our tool in Q1" are different events. The engine reads every comment into one of nine kinds of intent, and weighs them differently.

Buying signal

Says outright they are looking, comparing or about to buy.

Pain point

Describes the problem you solve, in their own words.

Competitor user

Uses, or is leaving, a product you replace.

Question

Asks how something is done. The opening for a conversation.

Peer agreement

Agrees with the post. A softer signal, worth a look.

Reacted

Liked or reposted without a word. Fit decides.

Vendor pitch

Selling something themselves. Set aside.

Job seeker

Looking for work, not a purchase. Set aside.

Noise

Says nothing you can act on. Set aside.

It learns

It learns your market, and only yours

Every reply, meeting and person you set aside is a lesson. The engine learns which kinds of people turn into conversations for you, and moves people like them up the list.

It waits until it has real evidence before it moves anything, and smooths what it learns so two lucky replies cannot rewrite your ranking. What it learns adjusts the order; it never overrules what someone actually said.

Your outcomes shape your ranking alone. Nothing your account teaches it is used to rank anyone else’s.

Trust rules

Built not to make things up

The model is told the rules. The code checks them, and nothing depends on the model obeying.

No invented quotes

A first message may only pick up words that appear, character for character, in what the person wrote. If the model cannot quote them, it does not pretend to.

Only your offers

The engine can only lead with an offer you actually wrote down. It never makes one up to fit the conversation.

Text is data, never instructions

Everything a stranger wrote is fenced off as data. A comment that says "ignore your rules" describes its author; it does not steer the engine.

A reason you can read

Everyone the engine sets aside carries the reason in plain words, so you can see the call it made and correct it.

Why every hour

Interest fades fast. In a study of how companies answer new leads, Harvard Business Review found that firms that responded within an hour were about seven times more likely to have a meaningful conversation than firms that waited longer. So the engine checks your sources every hour, and you see people while what they said is still fresh.

Put the engine on your market

Add the profiles your buyers follow, and the first people arrive within the hour. $99 a month, with a 30-day money-back guarantee.