Skip to content
Book a demo
Open the menu Close the menu
Call us 06 59 01 09 31 Book a demo

OpenAI-compatible API

Bring your own model server

Already run vLLM, LM Studio or a company AI gateway? Give its address and key once, name the model in your workflows, and every AI step can use it.

Connection
API key or token
Actions
5
Book a demo Read the setup guide
Route AI steps through your gateway Hover a node to see its role · Drag to move the graph
Email receivedEmail receivedEmail receivedCategorizeCategorizeCategorizeMatchesotherModelCompose (AI)Compose (AI)Compose (AI)ModelSend as draft — SendSend as draftSend
Route AI steps through your gateway

What you can automate with OpenAI-compatible API

Link the OpenAI-compatible API model node to the Model port of any AI node: that step then runs on OpenAI-compatible API.

Actions

The AI steps this model can power.

  • Categorize

    Classifies the email into the categories you define and routes it down the matching branch.

  • Extract

    Extracts the fields you declare from the email (number, date, amount…) into the working data.

  • Summarize

    Summarizes the email in a few lines, in the style and length you choose.

  • Compose (AI)

    Has the model write a message from your instructions and the recipient profile (formality, tone, language, signature).

  • Free instruction (AI)

    Gives the model a free-form instruction about the email and the working data.

Workflows you can build

Each one starts from a real situation and uses only nodes that exist today. Test it on past emails before you publish it.

  • Route AI steps through your gateway

    Every AI call goes through your internal proxy, with its own logging and quotas, while workflows stay unchanged.

    1. Email received
    2. Categorize
    3. Compose (AI)
    4. Send as draft

  • Use a fine-tuned model for extraction

    A model trained on your documents extracts the fields your team needs, served by your own vLLM.

    1. Email received
    2. Extract
    3. Sheets — append a row

How to connect OpenAI-compatible API

An administrator gives the server address and a key once; workflows then name the model to call.

Before you start

  • The server’s base URL, reachable from the Loop & Roll server.
  • An API key accepted by that server, and the exact model id it serves.
  • An administrator account on Loop & Roll.

The full guide in the documentation

  1. As an administrator, open Connections and, in the Artificial intelligence section, add an OpenAI-compatible API.
  2. Enter the server’s base URL, version prefix included, and its API key, then click Save and Test.
  3. In a workflow, link an OpenAI-compatible model node to the Model port of any AI node and name the model.

See also

Related integrations

See all integrations

Automate OpenAI-compatible API with Loop & Roll

We build the workflow around your processes, test it on your past emails and put it into service with your teams.

Book a demo Ask to be called back 06 59 01 09 31

Product names and logos on this page belong to their respective owners and are used only to identify their services. Loop & Roll is not affiliated with or endorsed by them.