Private beta · 30 free requests

Decide exactly what the model sees.

Yokolo is an AI workspace built around one idea: the transcript is not the context. Tag the messages and snippets that matter, switch the rest out, and run one prompt across several models.

Start typing — no account See it live

30 free requests · no card · your chats survive signup

yokolo.voidfl.com

Tag → exclude → compare

Live, not a video

The demo is the product.

This is the real app, running in guest mode. Type a question, then try tagging a message and switching it out of context.

yokolo.voidfl.com/app

Guest workspaces are isolated. Create an account when you want to keep one.

Why it feels different

Context is a choice, not a scroll position.

Most chat tools send the whole thread and hope. Yokolo lets you assemble the prompt deliberately, then compare what different models do with the same material.

Tag whole messages or just a snippet

Mark a decision, a constraint, a quote. Tags are labels you can act on later, not decoration.

Switch a tag out of context

Keep the transcript intact, but exclude what no longer matters. The model sees less, and you see whether it still gets the answer right.

One prompt, several models

Send the same context to Pro and Flash, or to different providers. Answers arrive side by side, labelled and preserved.

Your workspace is yours

Every account gets its own database. Bring your own provider key, or use the hosted trial while it lasts.

The core move

The transcript stays. The context changes.

A long thread is a record of everything you thought. A prompt should be the subset that bears on the question. Yokolo separates the two, so you can keep the history without paying for it on every request.

Tag decisions, constraints, quotes and examples as you go.
Toggle tags per prompt instead of editing the thread.
Anchor a context window to a range of text so edits don't break it.
See which model handles the reduced context best.

context, assembled

Comparing model answers with the same context

How it works

Three moves, then it clicks.

Write or paste

Start a conversation the way you normally would. Nothing to configure, no account required for the first 30 requests.

Tag what matters

Tag a whole message or highlight the sentence that carries the decision. Add a few tags and the context panel becomes useful.

Run and compare

Choose the tags in scope, pick two models, and send. The answers land side by side so the difference is visible.

Private beta

Try it before you decide.

No card, no install. Guests get 30 requests. Create an account and the workspace you already started comes with you — 250 requests a month while the beta is open.

Start free Create an account

Guest requests30 no account

Beta allowance250 requests / month

Beta size250 accounts

Questions

A few honest answers.

What counts as a request?

One message that asks the model for a response. Regenerating or sending to a second model counts as another request. During the beta we count requests rather than tokens so the limit is predictable.

Do I need an account to try it?

No. You get 30 requests in an isolated guest workspace. If you create an account later, that same workspace is upgraded in place, so the chats you started are still there.

Which models can I use?

The hosted beta offers Xiaomi MiMo Pro and Flash. Yokolo is provider-agnostic; a signed-in account can also bring its own provider key.

Is my data used for training?

No. Your conversations live in your own account database. Requests go to the model provider you use; the hosted beta uses a platform key and does not train on your data.

Why is the beta capped at 250 accounts?

It keeps the experience fast and the support honest while the product is still changing. The guest workspace stays open even when signups are full.