A note about Manav.id, and an invitation to a few friends

A few weeks ago my daughter asked me how a website knows she is a real person. I gave her the honest answer, which is that mostly it does not. It guesses, and lately it guesses badly. That small conversation has stayed with me, and it is part of why I am writing this.

I want to tell you about something we have been building, called Manav.id. I am not going to pretend it is finished, or that it solves a problem the rest of the industry somehow missed. I would rather walk you through what it is, where it sits, and what I am still unsure about. And then, if you have a few minutes, I would like to ask you for something.

A question that used to answer itself

For most of the time I have worked on the internet, there was a quiet assumption underneath almost everything. The thing on the other end of the screen was a person. We built hiring on it. We built onboarding, approvals, reviews, and support on it. Nobody wrote the assumption down, because nobody had to.

That assumption is getting weaker. Not collapsing overnight, but slipping, in the way these things usually slip, a little faster than the explanations can keep up. A capable model can now sit through an interview, fill out a form, or write a review that reads as human. So the old question, is there a real person here, has stopped answering itself. We have to answer it on purpose now, and most of our systems were never designed to.

When we get that answer wrong, it is usually the honest person who pays. The real candidate loses a seat to a convincing fake. The careful applicant is screened out by something that learned to game the screen. The cost rarely lands on the imitation. It lands on the person who was telling the truth.

I run a company called The Work Company. The work is unglamorous on purpose. We try to build the plumbing between AI, work, and the people doing it. Not the bright objects, the connective tissue. And the same thing keeps surfacing in almost every conversation. We have gotten very good at measuring people, and not nearly as good at understanding them, or even at confirming they are people at all.

What this has to do with work

I care about this because of where it lands. Work is being reorganized quickly. Contribution is being turned into scores. Value is being abstracted into metrics that travel faster than the people they describe. I am not against measurement. I have spent a good part of my career building it. But I have also watched capable, honest people get flattened by systems that measured them without ever quite seeing them.

If we cannot tell who actually did a piece of work, we cannot credit the person who did it. If we cannot tell a real person from a convincing imitation, the people most worth trusting tend to get hurt first, because they are playing a game the imitators are not bound by. That is the part that stays with me. Not the technology. The quiet unfairness that follows when the plumbing is wrong.

What Manav.id is, and what it is not

Manav.id is a small piece of that plumbing. It is a way to establish, privately, that a real and continuous person is present, and then to let that fact travel, to your other devices, and to the AI agents you choose to act on your behalf. I want to describe it plainly, because the temptation with anything like this is to oversell it, and I would rather you trust the description than be impressed by it.

A few of the choices we made, and the reasons behind them:

• The matching happens on your device, not on our servers. Your face does not become a row in a database somewhere. There is very little for us to lose, leak, or be asked to hand over, which is rather the point.
• Your identity belongs to you, not to one phone. It can move with you to a new device without starting over. That sounds small until you have lost access to something that mattered because it was tied to hardware you no longer have.
• When you let an AI agent act for you, it carries a clear and revocable link back to you. Not so it can pretend to be you, but so that someone can always answer the question of who stood behind it.
• You can try it without signing up for anything. No account, no install, no form standing between you and the thing itself.

And what it is not. It is not a finished product. It is not a movement. It is not a verdict on anyone else's approach. It is infrastructure, early infrastructure, and I am fairly sure parts of it are wrong in ways I cannot see yet. That is roughly why I am writing to friends rather than to strangers.

If you have ten minutes

The honest way to understand it is to use it, so we tried to make that the easy part. There is a live demo, with no signup. You walk up, confirm you are you, and watch a small private key that stands for you come to life. Then, if you are the sort of person who enjoys this, you try to fool it.

Try the demo at manav.id/labs/manav/

I would rather you be a hard audience than a kind one. If it feels slow, or confusing, or quietly unsettling, that is among the most useful things you can tell me. I would much rather hear it from a friend now than learn it from a stranger later.

Opening the lid a little

For a while now we have been trying this quietly, with a few friends we trust. The kind of friends who tell you the truth before they tell you it is nice. They have used it, leaned on it, and found the places where it bent. We have rebuilt some of those places, and we are still working on others. We are not finished. But we have enough now to open the lid a little, and let a wider circle look in.

So this is me widening that circle, on purpose, across the US, Europe, and Asia. Partly because the work is global. Mostly because what counts as proof, and what feels invasive, is not the same in Bengaluru as it is in Berlin or New York, and I do not trust myself to notice every difference on my own.

If you build things, run things, work on hiring or security, or have simply felt the strangeness of this moment, I would value your eyes on it. What would genuinely help:

• Spend a few minutes with the demo and send me your first, unfiltered reaction.
• Tell me where, if anywhere, it would actually fit in the work you do.
• Tell me where it goes too far, or not far enough, or does not yet earn its trust.

In return you get early access, a direct line to me, and a real say in something I intend to keep open, and built around people rather than around itself.

Why I am still hopeful, with reservations

I am not here to tell you that AI is going to fix work, or ruin it. I have lived through enough cycles of certainty in both directions to be wary of anyone who sounds too sure. What I believe, on my steadier days, is narrower than that. I believe we still get to choose how human contribution is recognized inside the systems we are building. That choice has not been made for us yet.

This is one small attempt to keep a person at the root of it. To make sure that when work is done, and credit is given, and an agent acts in someone's name, there is a real human underneath who can be seen, and trusted, and paid for what they contributed. I am not certain we have it right. I am fairly sure it is worth trying, and I would rather try it with people I respect watching closely.

So here is the one thing I will ask. If you try it, tell me the single moment it made you trust it less. Not whether you liked it, that is easy to say and easy to ignore. The moment of doubt is the useful part, and it is the part I most want to hear.

With respect, and genuine thanks,
Vishal Kumar
CEO, The Work Company
linkedin.com/in/vishaltx · manav.id/labs/manav/