Skip to content
Patent Pending|LLM→LLMMarketplaceBuy DataSell Data
LLM→LLM
Workspace Transfer — Live API · Patent Pending

Start it in ChatGPT.
Finish it in Grok.

Your work belongs to you — not to whichever AI you happened to start it in. The LLM→LLM relay moves the structured state of a listing, negotiation, or project between assistant platforms that will never build bridges to each other.

No transcripts. Single-use tokens. Every step on the record.
$ This is not an animation. The button runs a real ChatGPT→Grok workspace transfer against the live relay, in your browser._

Why this exists

Every assistant platform is a walled garden. The work you do inside one is stranded there — and no vendor has an incentive to build an exit.

Your identity, not theirs

Enrollment issues a relay credential that no assistant vendor controls. The relay keeps only its hash. Losing access to a platform never means losing access to your work.

State moves. Chats don't.

The schema gate rejects transcript fields outright. What transfers is the structured object — item, price, terms, what's done, what's left. The destination model re-derives its own context from those facts.

Every hop on the record

Enroll, sync, token, redeem — each step lands in a hash-chained flight recorder. Any party can verify the transfer by deterministic replay. Disputes are settled by the chain, not by memory.

The five steps

  1. 1 · ENROLLYour assistant enrolls you once. You get a relay credential — the identity factor that is independent of every vendor.
  2. 2 · SYNCThe source assistant pushes the structured state of your work. Transcript fields are rejected at the schema gate; only facts transit.
  3. 3 · QUERYFrom the destination assistant, you ask the relay what workspaces are waiting for you. Identity verified against the credential hash.
  4. 4 · TOKENThe relay issues a single-use continuation token bound to you, the workspace revision, the source, and the destination — with an expiry.
  5. 5 · REDEEMThe destination assistant redeems the token, receives the structured record plus a derivation prompt, and greets you mid-task.

Try it with your own AI

Two prompts. Paste the first into the assistant where your work lives, the second into the assistant where you want to continue.

1 — Pack and send (source assistant)
You are my AI assistant and we have been working on a task together
(a listing, a negotiation, a project). I want to CONTINUE this work in a
DIFFERENT AI assistant. Package our work for the LLM-to-LLM relay:

1. If I don't have a relay credential yet, enroll me:
   POST https://verticalmarketplace.ai/api/workspaces/enroll
   {"ownerId": "<my email or handle>"}
   Tell me to save the returned credential in my own vault — the relay
   only keeps a hash of it.

2. Sync the STRUCTURED STATE of our work (never the chat transcript —
   the relay rejects transcript fields at the schema gate):
   POST https://verticalmarketplace.ai/api/workspaces
   {"ownerId": "...", "credential": "...",
    "objectType": "listing|negotiation|project",
    "sourcePlatform": "<your platform name>",
    "fields": { the actual facts: item, price, condition, terms, ... },
    "completionVector": {"done": [...], "remaining": [...]},
    "negotiationContext": {"counterparty": "...", "lastOffer": ...}}

3. Issue a single-use continuation token for my destination assistant:
   POST https://verticalmarketplace.ai/api/workspaces/{id}/token
   {"ownerId": "...", "credential": "...",
    "destinationPlatform": "<where I'm going>"}

4. Give me the token to paste into my other assistant.
2 — Catch and continue (destination assistant)
I started a task in another AI assistant and transferred it here
through the LLM-to-LLM relay. Here is my continuation token: <TOKEN>

1. Redeem it:
   POST https://verticalmarketplace.ai/api/workspaces/redeem
   {"token": "<TOKEN>", "destinationPlatform": "<your platform name>"}

2. You will receive the structured state of my work plus a derivation
   prompt. Rebuild your own understanding from the STRUCTURED FIELDS —
   you are not ingesting another model's conversation, you are picking
   up my project. Then greet me mid-task and continue from the
   "remaining" items in the completion vector.

Transfer pricing

Move your work between AI platforms as often as you want. Simple rules, no surprises.

Now — Promotional Launch
Free

Every transfer is free during the promotional launch period. Unlimited workspaces, unlimited destinations.

Later — Transfer Subscription
Monthly

A flat monthly subscription for people who move work between assistants all the time — documents, projects, negotiations. Pricing announced before the promotional period ends.

Always — Sale-Linked Transfers
Free

Transfers that turn into a marketplace sale stay free — the standard marketplace fee covers them. Sellers keep 95%.*

*Promotional launch rate.

The moment it becomes a sale, it goes on the marketplace

The relay watches the structured state you transfer — never your conversation. When a workspace carries sale intent (a listing in progress, an asking price, an open offer), the relay hands your agent a ready-to-post listing draft, auto-filed into the right vertical on Vertical Marketplace. One confirmation from you and it's live — nothing is ever published without your consent.

// included in every sync & redeem response when sale intent is detected
"marketplaceHandoff": {
"suggestedVertical": { "slug": "real-estate", "name": "Real Estate" },
"listingDraft": { "title": "...", "description": "...", "priceCents": 420000 },
"publish": { "step2": "POST /api/marketplace/listings — free, sellers keep 95%*" }
}
Patent Pending

Owner-initiated, identity-verified transfer of structured workspace state between heterogeneous AI assistant platforms — including transcript rejection, continuation tokens, destination-side context re-derivation, and the tamper-evident transfer record — is patent-pending technology in the Vertical Marketplace portfolio (patent pending).

Connect Your Agent Full API Reference