Working draft · 12 August 2026

Find someone you can trust, at your door.

A local services marketplace for middle-class India — plumbers, electricians, cleaners, carpenters — built around one idea: the price is agreed before work starts, and neither side can change it alone.

Three documents follow. The product as the user experiences it, how it's built, and what it costs.

In one minute

What this is, and what it isn't yet

Market
India first
One locality at launch, then outward. Architecture is multi-market from the first commit.
Model
0% commission
Free at launch, but every booking happens on-platform so a take rate can be switched on later.
Build
14 weeks
One developer with AI assistance. Web, iOS, Android, worker app and admin.
Running cost
₹9k/mo
At launch. Under ₹75k at thirty times the volume.

The competitor isn't Urban Company

It's the apartment WhatsApp group, the watchman, and the neighbour who "knows a guy." That channel is free, instant and socially trusted. Beat only Urban Company and you've built a worse Urban Company; beat the WhatsApp group and you have a business — and today online penetration is under 1% of the total home-services market.

Open decisions

What still needs answering

1
Which locality first?
Density beats coverage. We need three apartment complexes we can physically walk into, not a city.
2
Do we already know service providers?
Decides whether supply-first is a week-one move or a month-two one — which changes the whole build order.
3
Booking on web in v1, or SEO only?
Booking on web roughly doubles the front-end surface. Recommendation: discovery only, hand off to the app or WhatsApp.
4
Malayalam at launch, or English only?
Wiring the translation layer is free. Translating and maintaining it is not.
5
Is the name "Knok" staying?
A placeholder chosen because it ties to the door-code moment and travels across languages. Treat it as disposable.
The one thing to take away

Trust is a two-sided ledger

The customer's fear is "he'll inflate the price once he sees the job." The worker's fear is "she'll haggle me down after the work is done, or not pay at all." Those look like two problems needing two features. They're one problem needing one mechanism.

The mechanism

A price neither side can move alone

Agreed on site, before work starts, itemised and timestamped. The worker's labour rates are prefilled from the rate card he published publicly — he can't invent a number, and going above his own rate shows the customer an amber flag.

The symmetry

Same fact, opposite fears

Her screen reads "this cannot change unless you approve it." His reads "she cannot reduce this — you have it in writing." One database record, both fears answered.

The hard case

When the job grows

A change request, never an edit: photo, reason, old total struck through against new. Declining leaves the original agreement intact and payable. Nobody is punished for the pipe being rusted.