Skip to content
Private preview

Keep the work.Keep the context.

Okroh is a local-first AI workspace for conversations, projects, documents, memory, and artifacts—held together in one desktop surface.

Coming Soon See the workspace

Coming soon to macOS, Windows, Linux, iOS, and Android.

On your devicesChats, files, and learned memory stay local.
Across your devicesEncrypted updates carry memory between them.
Model-ready contextSwitch models without scattering the thread.
Workspace surfaceContext, in view
your work
stays close
memory moves
when you do

The useful parts of AI, in one calm surface.

A sidebar for what you are carrying, a reading surface for the conversation, and a composer that keeps the next move obvious.

Explore the workspace guide
okroh-demo.mp4
muted loop
Preview unavailable The workspace still makes sense without the loop. Open the product preview video
sound off · interaction optional
Conversation → context → artifactA short loop of the product in use. Open video
SurfaceWorkspace
PlaybackMuted loop
FormatMP4 · 16:9

Local working set

Conversations, documents, and learned context stay close to the device doing the work.

Ephemeral handoff

When continuity moves between devices, the relay is a passage—not a place where your work lives.

Recoverable by design

Authorized devices keep the path back to your workspace in your hands.

A workspace that keeps the thread.

Okroh makes the parts around a conversation useful: project context, local memory, connected files, and the artifact you are making together.

Artifacts · open beside the thread

The answer can become something you can use.

When a conversation turns into a plan, chart, document, or interactive widget, Okroh gives it a surface of its own—so the output stays close to the reasoning that made it.

canvas / live

Project context

Bring local files into the conversation without losing the thread.

local

files ↔ conversation

Learned memory

Review, correct, or forget the preferences Okroh carries forward.

yours

reviewable / yours

Connectors and skills

Choose which tools can help in this conversation, when they matter.

on purpose

chosen tools / clear boundary

Step-by-step work

Complex tasks can stay legible while Okroh searches, checks, and synthesizes.

clear

search → check → synthesize

The boundary is part of the product.

Okroh makes the path of your data visible: what stays on your devices, what passes through a relay, and what a model needs only long enough to answer.

Read the privacy architecture
A request, made visible.
  1. 01 / device stays on your devices

    Conversations, documents, and learned memory

    Stored locally so your workspace can stay useful without making your working context a server-side archive.

  2. 02 / relay passes through

    Sync packets move through an ephemeral handoff

    When you continue on another device, encrypted updates pass through the relay and are not kept as a readable history.

  3. 03 / answer needed to answer

    Models get what they need to respond

    Okroh routes prompts and context through zero-retention enterprise agreements, then brings the response back into your workspace.

The difference is where the context lives.

Okroh keeps the workspace around the conversation, instead of asking you to rebuild the same context for every model or device.

Where it stays
Standard AIone destination
Okrohone workspace
Your data
Your chats are saved on the company's servers.
Your chats, files, and memory stay on your devices.
Memory
Memory is stuck inside one app and one AI model.
One local memory can support the models and devices you choose.
AI models
You only get the models that one company makes.
Switch between leading models in the same workspace.
Subscriptions
A separate plan for every AI service you use.
One Okroh subscription covers every model in the app.

The payoff: the conversation, the source material, and the next move can stay in the same place.

One memory, wherever you use AI.

Tell Okroh something once, then let the right model pick up the work on the device you choose.

Learned memory is reviewable: you can correct it, remove it, or see what is no longer current. Your account manages access and billing; your working context stays on your devices.

How key recovery works
One memory / many surfacesencrypted continuity
Okroh Private preview in progress

Make room for the work.

Okroh is launching soon on macOS, Windows, Linux, iOS, and Android. Bring the conversation, the context, and the next step with you.

Coming Soon
One workspace, carried forward.
  • The conversation stays understandable
  • The context stays close to the work
  • The next move stays obvious