UKI

Your AI learns from you overnight

The Uki owl: one eye closed, one eye open
One eye sleeps, the other learns.

Uki lives on your phone. While it charges, the app turns your notes, chats and documents into training examples and fine-tunes a small open model on them. By morning it knows your work better and writes the way you do — and your data never leaves the device.

  • iPhone and Android from day one
  • Your data never leaves the phone
  • Speaks Kazakh and Russian
  • Works offline

Today an AI that really knows you is either someone else’s or built for engineers.

Cloud assistants are someone else’s

One model shared by everyone. Your personal data goes to someone else’s servers, and personalisation stops at “memory” and instructions.

Fine-tuning is for engineers

Even the best open tools need a graphics card, a command line and a ready-made dataset.

The real barrier is data, not hardware

Nobody knows how to turn their chats and notes into training examples — or how to check that the model actually got better.

But everyone has a phone

Most people don’t own a computer with a graphics card. Their phone has a powerful GPU — and it sits idle every night.

One night with Uki

Every night your phone turns your data into your own AI. The cycle repeats on new data.

All you do — three actions
  1. 01 Connect your sources
  2. 02 Put the phone on charge
  3. 03 In the morning, compare before / after
  1. daytime You connect your sources A Telegram export, notes, documents, voice messages. Once.
  2. 23:40 Phone goes on charge Uki gets to work. Nothing more is needed from you.
  3. 23:45 Clean-up Passwords, card numbers and other people’s contacts are removed automatically.
  4. 00:30 Auto-dataset Your texts become question-and-answer pairs and samples of your style.
  5. 01:00 Night training A small LoRA adapter is trained on the phone’s GPU on top of an open model.
  6. 06:00 Check The model answers your personal questions and runs general tests. If it got worse, the night is rolled back.
  7. 07:00 Your AI in the morning Check passed: it writes like you and knows your work. You see the before / after.

* Times are illustrative. Our target: a night of training fits within 6 hours on charge, without overheating.

See the difference by morning

The same request, before and after one night of training on your own messages.

request: “Tell the client I’ll deliver the mockups on Friday”
before — a generic model

Dear Client, I am writing to inform you that the mockups will be delivered on Friday. Thank you for your understanding. Best regards.

after — your model

Hi Aigerim! You’ll have the mockups on Friday before lunch — I’ll send a Figma link, same as last time.

We measure, not promise. Every model goes through a blind before / after comparison. Our goal: people pick the trained model in at least 70% of cases.

And from day one search across your own data (RAG) already works — before the first night of training.

Your model.
Your data.
Your phone.

Everything stays on the device

Training and inference run on the phone. Your data is not sent anywhere.

Open-source core

Anyone can check what happens to the data.

Offline mode

The model works and learns even without internet.

Our vision

Training your own AI should be as easy as making your own website. Own your model — don’t rent someone else’s.

Who Uki is for
We start here

Freelancers and writers

Writes client emails and posts the way you write them. You can see the difference with your own eyes.

Students

Knows your notes, lectures and course materials.

Anyone who values privacy

An AI of your own that sends nothing to the cloud.

Start for free

Training and inference run on your phone, so our server costs are close to zero — and the basic version is free. Prices are hypotheses we’ll confirm in the beta.

Free

One model, search across your data, training once a week.

$0

Pro

Nightly training every day, all data sources, several “personas” — for work and for study.

$5–8 / mo

Teams

A shared team model, trained together on employees’ devices; data never leaves the company. Planned for year two.

per user · later

Later: a catalogue of ready-made adapters — trained models that people share or sell.

Roadmap

Four stages in the first year. Each next stage starts only once the previous one passes its threshold.

  1. 0 weeks 1–4 · now

    Idea check

    Auto-dataset, blind test, 10 interviews, background-training prototype.

    threshold → ≥70% pick the new model in a blind test

  2. 1 months 2–3

    Phone

    Training on iPhone and Android, measurements, hiring ML and mobile engineers.

    threshold → a night ≤6 h without overheating

  3. 2 months 4–6

    MVP and beta

    The app, data import, the before / after screen, 100 beta testers.

    threshold → 40% come back after a week

  4. 3 months 7–12

    Launch and growth

    App Store and Google Play, Pro subscription.

    threshold →

Questions

Which phones will it run on?

iPhone and Android, from the first version. Flagships train the model right on the phone; other phones get search across your data plus training on your laptop over Wi-Fi, or a smaller model.

Will my phone overheat?

Training only runs while the phone is charging and you are asleep. Our threshold: a night of training within 6 hours, no overheating. Heat and battery are the first things we measure.

Will the model get worse at everything else?

After every night the model takes a set of general tests. If something drops, the training is rolled back.

What data can I connect?

Telegram exports, notes, documents and voice messages. Passwords, card numbers and other people’s contacts are removed before training.

Do I need internet?

No. There is an offline mode, and your data never leaves the phone either way.

Plug it in.
Wake up to your own AI.

Join the beta: we’re picking the first 100 beta testers and will ask what your AI should be like. Starting in Kazakhstan and the CIS, on iPhone and Android.