Skip to content

Section: AI & Privacy

AI Features

What Kuro AI is meant to be, how it is designed to work, and what is not yet confirmed.

What Kuro AI is

Kuro AI is an optional system. The design is for you to choose whether to enable it during onboarding, and to be able to change that choice later in Settings. It is built as a separate module so that turning it off does not affect the rest of the browser.

Local-first by design

The project's priority is to process things locally where possible. The planned architecture sends requests through a local interface to Ollama, which runs a model on your own computer.

  1. Kuro

    The browser

  2. Local AI interface

    Kuro's connection to the model runner

  3. Ollama

    Runs models on your computer

  4. Local model

    Generates the responses

Planned architecture of Kuro AI

Capabilities the project is exploring

None of these should be assumed to be available yet.

  • Page understanding

    Help making sense of the page you are viewing.

    Planned: On the roadmap. Not available yet.
  • Selected-text analysis

    Explain or analyze text you select.

    Planned: On the roadmap. Not available yet.
  • Summarization

    Summaries of pages and other content.

    Planned: On the roadmap. Not available yet.
  • Research assistance

    Help gathering and comparing information.

    Planned: On the roadmap. Not available yet.
  • Tab organization

    Help sorting and organizing open tabs.

    Planned: On the roadmap. Not available yet.
  • History assistance

    Help finding things you have visited before.

    Planned: On the roadmap. Not available yet.
  • Writing assistance

    Help drafting and editing text.

    Planned: On the roadmap. Not available yet.
  • Local browser search

    Searching browser content using local processing.

    Planned: On the roadmap. Not available yet.
  • Context-aware assistance

    Assistance that takes what you are doing into account.

    Planned: On the roadmap. Not available yet.
  • Browser automation

    Having the assistant carry out browser tasks for you.

    Planned: On the roadmap. Not available yet.

Current status

Early prototype work has used a small local model through Ollama. There is no list of supported models yet and no user-facing setup guide.