KADARI

What we do with
what you share.

Plainly: the only thing this site collects is a work email, and only if you type it in. We don't sell it, we don't track you, and you can have it deleted by asking. Last updated 9 August 2026.

What we collect

One thing: the work email addressyou enter in the partner form. There are no other fields. We don't use tracking cookies, analytics, advertising trackers, or session recording on this site — the pages you are reading are static files.

Where it goes

The form posts to Basin (usebasin.com), a form provider with EU data residency, which delivers the submission to us by email. That is the only third party involved, and the only information it receives is what you typed.

Why we collect it

Only to reply to you about becoming a design partner. We do not sell it, rent it, use it for advertising, or share it with anyone outside Kadari.

How long we keep it, and how to have it deleted

We keep your message for as long as our conversation is live, and delete it on request. Email [email protected] at any time to see what we hold about you, correct it, or have it removed. No hoops, no form.

The client, and your prompts

The kadari client is local-first. It opens no network connection, calls no model itself, and uploads nothing: it appends a record of calls you already make to a log file on your own machine, at a path you choose.

One thing worth knowing if you are running it today: in 0.3.0 and earlier, that log is created with your system's default file permissions, which on most machines means other accounts on the same host can read it. If you share the machine, check the file's permissions — ls -l on the log will tell you, and it costs nothing to look.

0.3.1 is the release that changes it: the log, and everything else the tool writes, is created readable only by your own user. Rather than take a date from this page, check for yourself — kadari version tells you what you are running, and if pip install --upgrade kadari offers 0.3.1 or later, take it.

There is no upload endpoint on this site. Nothing here can receive a log, automatically or otherwise. If you want an evaluation today, you package a log yourself and we arrange the hand-off personally.

What this page does not cover yet

It covers this website and the client. It does not yet cover handing us a capture log for evaluation — where such a log would be stored, who could read it, how long it would be kept, and how it would be destroyed.

Those terms are not written, so this page does not pretend to them. They will be published here before any path exists to send us a log, not after. If you are evaluating us and need them sooner, ask and we will write them with you.

If we ever host anything

A hosted tier, if we offer one, would be opt-in, isolated per customer, and aggregate-only: one customer's data never enters another's path, and raw prompts and outputs never accumulate into anything shared.