The law belongs to 240 million people. Its language belongs to 250,000.

The problem

Almost nobody can reach the law.

It is not a literacy problem. A statute is drafted in a register built for courts, not citizens — and no one can tell you which of 5,325 Acts governs the thing that just happened to them. So most people never ask. They absorb it.

What we already are

We know, because we have been doing this by hand.

TVL is not a deck. A 162,000-member community brings us their problems; people reach us on WhatsApp; we work the query out and connect them to a real lawyer — from a network that already spans Pakistan. All of it by hand, today. The engine does not create that demand or that supply. It carries what we are already carrying.

Most startups raise to find demand and supply. We raise to serve the demand and supply we already have.

Two things kill consumer marketplaces: nobody comes, and nobody answers. Both are already solved here, and neither was bought. What we run today — reach, intake, triage, referral — is the marketplace; it just runs on human hours, so it stops when we sleep and it cannot serve 240 million. That is the thesis: an automation bet, not a market-creation bet. Every question a volunteer answers by hand tonight is one the engine should have answered in nine seconds, cited, for free, at 3am, in Sialkot — leaving our lawyers for the cases that actually need a lawyer.

What we already sell

And some of them already pay us.

The hardest question asked of any consumer-legal company in Pakistan is whether ordinary people will ever pay for legal answers. We are not going to argue it from a survey. We bank it, in three lines, every month — all of them fulfilled by hand.

Our customers are already buying the engine’s output. They are just buying it at the price of a researcher’s afternoon.

This is why the raise is not a demand experiment. The order already comes in; the only question is what fulfils it. Today it is hours — so the service sleeps, it queues, it costs a wage every single time, and it can never reach 240 million. With the engine the same order is fulfilled by retrieval and generation: seconds instead of days, concurrent instead of one-at-a-time, and a marginal cost of electricity instead of salary, because the model is open-weight and runs on our own hardware. That is not a new revenue line. It is the same revenue line, with the labour taken out of it.

And the exhaust is the asset. Every paid citation request is a real question from a paying customer, answered with a verified authority by our own researchers. That is a labelled dataset of Pakistani law in the exact distribution real people ask it in — the benchmark and the training corpus a domain model would need, accumulated as a by-product of getting paid.

Said plainly: these lines are small, manual, and founder-reported until we instrument them. They prove that people will pay. They do not prove how many. We keep those two claims apart.

The market’s blind spot

So 38 startups looked at this — and all of them built for the same 0.1%.

Roughly 38 Pakistani legaltech startups exist; only about four are funded. Almost every one is English-first and lawyer-first. They are competing for the sliver.

The honest competitor isn’t any of them — it’s “just ask ChatGPT.” Which answers a Pakistani tenancy question fluently, confidently, and with a citation that does not exist. In law, that is not a competitor. It’s a liability.