How to Use Jev in ChatGPT: TypeSafe's Decision Model via MCP

Jev in ChatGPT

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

Short answer: install the Jev connector on AnythingMCP with your TypeSafe API key, add AnythingMCP to ChatGPT as a custom MCP server (Plugins, +, Add custom MCP server) with the URL https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID. ChatGPT can then pass every yes/no, classification or scoring decision to Jev and gets a probability back instead of a guess.

Jev, from TypeSafe, is a decision model: it never writes text, it answers typed questions about the evidence you give it. That makes it fast (about 300 ms), cheap (input tokens only) and explicit about uncertainty. ChatGPT stays in charge of the conversation.

Set it up

Step 1: Install Jev on AnythingMCP

1. Create an API key at console.typesafe.ai.
2. Create a workspace on AnythingMCP Cloud (7 days free) or use your self-hosted AnythingMCP.
3. Open Connectors, find Jev by TypeSafe, click Install and paste the key into TYPESAFE_API_KEY.

Step 2: Add AnythingMCP to ChatGPT

ChatGPT adds MCP servers as plugins on its Plugins page, on the web and on paid plans. Developer mode is no longer needed. In a Business workspace only admins and owners can add one; in Enterprise and Edu an admin can grant the Create plugins with MCPs permission.

1. Go to chatgpt.com/plugins, click +, then Add custom MCP server.
2. Enter a name, e.g. Name: AnythingMCP.
3. Under Connection, enter the Server URL https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID and keep Authentication on OAuth.
4. Select I understand and want to continue, then Create as a plugin. When ChatGPT asks, sign in to AnythingMCP and allow access.
5. Install the new plugin from your personal plugins, then type @ in a chat and pick it.

Step 3: Ask ChatGPT to use Jev

Start a new chat, pick the AnythingMCP app and mention Jev, for example "use Jev to rate these leads". ChatGPT reads jev_playbook and calls the Jev tools on its own.

What happens when you ask

Ask ChatGPT to route twenty support tickets to billing, technical, sales or none. It writes the four options with a short description each, calls jev_classify once per ticket with the ticket text as evidence, and gets back the chosen team, a probability for every team and a confidence. It can then show you a table and list separately the tickets where the confidence is low, so a person looks at those.

Prompts to try in ChatGPT

• "Use Jev to classify these tickets into billing, technical, sales or none and show the confidence for each."
• "Ask Jev which of these customer messages mention a cancellation, and give me the probability for each."
• "Rate these leads from 0 (no fit) to 3 (ideal) with Jev against the profile below."
• "Check with Jev whether each FAQ answer below actually answers its question."

FAQ

Does it work on the free ChatGPT plan? OpenAI documents custom MCP servers for paid plans only; the Free plan is not on its list.

What does it cost? Jev is billed by TypeSafe, about $0.042 per million input tokens, output free; a typical decision uses 300 to 520 tokens. AnythingMCP Cloud has a 7-day free trial, then from €19 a month; self-hosted AnythingMCP is free.

Why not just ask ChatGPT to decide? You can, but you get a sentence back. Jev returns a number on a fixed scale, the same way for every record, and tells you when it is unsure. That is what you need to sort, filter or automate.

Can I use the same setup in Codex or Claude? Yes. Every MCP client can connect to the same AnythingMCP server and use the same Jev connector and key.

Related

• Jev MCP server: the overview
• Connect Jev to Claude
• Triage support tickets with Jev in Claude
• Connect Jev to Cursor
• Connect Jev to Gemini CLI
• Connect Jev to Meta Muse
• Jev by TypeSafe in the marketplace