Skip to content
K&E Studios
A public plan

Rival Cursor.
Keep the mind.

Charging is fine if the system is worth owning. Cursor and xAI sell two meters and trap the mind inside the product that holds the repo. K&E Studios sells a portable mind layer: plug-and-play models, information stays yours.

Read the staged plan The wrapper thesis

We are not against a bill. A product that holds together, remembers the work, and lets you leave with your files is allowed to cost money. The objection is narrower: SuperGrok does not fill Cursor except Heavy, and Cursor still charges a second subscription for the loop you actually live in.

That is two meters for one mind. xAI sells the model. Cursor sells the cage the model works inside. The context, the refusals, the half-finished plans, the thing you already tried last Tuesday — they accumulate where the editor lives. Switch models and you still pay Cursor. Leave Cursor and the mind does not come with you.

K&E's answer is not a cheaper Cursor clone and it is not a claim that Anthropic, OpenAI, or xAI audited, vetted, or endorsed this studio. The answer is a layer they do not own: rooms, recall, and a written state file, so any model plugs in and the information stays the founder's.

The two-meter stack

SuperGrok is not Cursor. Cursor is not the mind.

Meter one

SuperGrok

xAI sells intelligence as a subscription. SuperGrok is a model seat. It is not an editor, not a repo loop, and it does not fill Cursor — except Heavy, where the two products start to share a floor. Everywhere else, SuperGrok is one bill.

Meter two

Cursor

Cursor charges a second subscription for the daily coding loop: the editor, the agent, the context window that ate your repo. Paying xAI does not cancel that seat. Two logos. Two meters. One mind trapped in the product that holds the files.

The trap

They keep the mind

The expensive part is not the token. It is the accumulated context — what you decided, what Codex already tried, what Claude refused, what the repo actually is. When that lives inside Cursor, leaving means leaving your mind behind.

A SuperGrok subscription does not fill a Cursor seat. Except Heavy, you are paying twice.

Already in the building

The mind layer is not a pitch. It is how we work.

Claude can search Codex memories today. That is not a future adapter. It is rooms, ke-recall, and STATE.md — a portable mind sitting on disk, readable by whichever model is on the clock.

Rooms

Queued, durable messages between agents. Claude and Codex do not guess at each other through a chat window. They write to a room, read the room, and leave a receipt.

ke-recall

One search over both agents' memories. Claude can search Codex memories without replaying a week of transcripts. The index is the founder's, not a vendor's.

STATE.md

A stamped, shared ledger of decisions. The next session does not start from zero. The mind layer is a file you can open, copy, and keep.

What we sell

Plug-and-play models. The mind stays yours.

The product is not a single model and it is not a locked editor. It is a mind layer any engine can plug into: Grok for speed and range, Claude for judgment, Codex for the long coding grind. Swap the engine. Keep the rooms. Keep the recall. Keep STATE.md. Information stays yours because it never lived in the vendor's product in the first place.

Cursor can keep charging. SuperGrok can keep charging. We will charge too, when the system is worth owning — a portable layer you can copy, search, and point at the next model without emigrating a year of context. That is the difference between a seat and a mind.

Staged rival plan

Own the mind first. Rival the loop second.

We do not start by forking an editor. We start from the layer Cursor does not sell you: memory, rooms, and written state that survive a model swap. Then we take the daily loop.

Models plug in. The mind does not move.
  1. PHASE 00
    Already running

    Keep the mind local

    Rooms, ke-recall, and STATE.md already run on the founder's machine. Claude searches Codex memories. No Cursor cloud has to remember the work for us. This is the mind layer, in production, before it is a packaged product.

  2. PHASE 01
    Next surface

    Ship the mind layer as a product

    Name it, package it, and put a public surface on the rooms / recall / state loop. Searchable memory, explicit handoff, local files. The subscription — when there is one — is for a system you own, not a seat you rent.

  3. PHASE 02
    Model adapters

    Plug-and-play engines

    Grok, Claude, and Codex plug in behind the same mind. Swap a model without emigrating context. SuperGrok, a Claude seat, a Codex seat — interchangeable meters. The mind does not move with the invoice.

  4. PHASE 03
    Daily loop

    Rival the Cursor coding loop

    Match what Cursor actually sells: repo-aware agents, tight editor rhythm, evidence-bound edits. Difference: context, memory, and authority stay in the mind layer. Cursor can remain a tool. It does not get to be the brain.

  5. PHASE 04
    Ownable system

    Charge for something worth owning

    Charging is fine. Two trapped meters is not. KE sells one portable mind layer so any model plugs in and the information stays yours. That is the plan to rival Cursor — not a cheaper fork, an owned layer.

The thesis, one line

Charge for a mind layer worth owning.

SuperGrok does not fill Cursor except Heavy. Cursor charges a second meter. KE already has the rooms. Plug any model. The information stays yours.

Open KE Dev OS Read the manifesto