Marco. The guy who built dateread.

I am Marco. I am a builder. I am not a relationship coach and I am not a guru. I have never run a workshop, sold a course, or posted a thread that started with men, listen. The thing I am is the guy who reread the same five messages on a Wednesday night trying to figure out whether what she said meant what I hoped it meant.

For years I assumed that was a me problem. The kind of private overthinking you keep to yourself because the advice you would get if you asked is either too generic to apply or too aggressive to use. Then I started noticing every friend I trusted was doing the same thing, just in a different group chat. Different girl, same loop. So I stopped treating it as a bug in me and started treating it as a thing that needed a tool.

The honest version is that there is a moment, usually around 1 a.m. on a weeknight, when you want one person you trust to read the thread and tell you what they see. Not their opinion of you. Not their opinion of her. Just the read. Most of the time at 1 a.m. nobody picks up. The friend you would ask is asleep, or busy, or you have already asked them about this same girl twice and you do not want to be the guy who is still asking.

dateread is the version of that friend who is always awake. Not because it is some replacement for the friend. Because it does the one part of the work that is mechanical. The part where you count what is actually in the messages, instead of feeling what you wish was in them. The friend you call at 1 a.m. is the one who tells you, plainly, what is on the page. That is what I built.

I tried a lot of other things before I landed on the four. Sentiment scores were a dead end. Tone classification was a dead end. Anything that tried to guess what she felt got it wrong in the way that hurts most, which is confidently. The four signals work because they are countable. You can point at them, you can tell me which message they changed at, and you can be wrong about the read but you cannot be wrong about the count.

Question asymmetry, latency drift, register shift, initiation flip. Four things that actually move in real threads, four things a friend would notice without calling them by these names. If you want the long version of why each one matters, I wrote it up over on the methodology page. The short version is here.

It is not flirty pickup lines. It is not negging, gaming, alpha framing, or any other shape the bad version of this category takes. It does not draft messages that treat her like a target to be moved. If a draft comes back that reads like that, the read failed and I want to know about it. Send me the thread.

It is also not a tool for winning. There is no winning a thread. The whole frame of winning is the thing that gets in the way of reading what is actually there. The tool exists so you stop spending your week guessing whether the silence means something. If the read says go warmer, go warmer. If it says go quiet, go quiet. If it says the answer is in the count and the count is against you, the move is to stop reaching. None of that is a script. All of it is the thing a friend would say if they were honest.

I also build tgkit, a small set of Telegram utilities. That is the rest of what I do publicly. dateread is the one I built for myself first, then for the friends who kept asking, and now for whoever ends up at 1 a.m. with the same five messages open.

The long version of the four signals is on the methodology page. The applied read, ten countable patterns in a cooling thread, is on the signs she's losing interest pillar. The named concept page that goes deepest on the timing signal is latency drift. The compact reference of all four lives on the glossary. Start with whichever one matches the thread you are rereading right now.