• DATA
  • LOGIC
  • ACTION
  • HUMAN

How to run machine learning inference with your AI

Connect your agent to Modelbit to turn raw data into model predictions instantly. Once connected, your AI handles the heavy lifting of model communication.

Ask AI about this page

Short answer

How can I run ML inference through my AI?

You send data from your AI agent to a Modelbit endpoint to receive immediate predictions. This allows your agent to use custom machine learning models as part of its reasoning process.

The results of inference

Where your data goes

What happens when your agent runs a model.

DATA

Instant predictions

The agent sends input to the model and gets back a prediction. You get the exact output your model was trained to provide.

LOGIC

Model-driven decisions

Your agent uses the model's output to decide what to do next. This turns a static prompt into a dynamic, model-aware workflow.

ACTION

Automated responses

The agent can take the prediction and immediately act on it. This moves you from just seeing data to executing tasks based on it.

HUMAN

Review required

You decide if a prediction is reliable enough to act on. The agent provides the data, but you keep the final say on high-stakes decisions.

The workflow

What your AI does when data arrives.

The agent manages the entire request-response loop with your model.

  1. Prepare input

    The agent formats your text or data to match the specific requirements of your deployed model.

    get_inference

  2. Execute inference

    The agent calls the Modelbit endpoint to run the model against the prepared data.

    get_inference

  3. Parse results

    The agent reads the raw prediction from the model and translates it into a format it can use.

    get_inference

  4. Apply prediction

    The agent uses the resulting prediction to update its context or trigger a specific response.

    get_inference

Try it

Copy these to start.

Use these prompts to trigger your first inference.

Starting points

Swap out the bracketed text with your actual model name or data type.

Modelbit Connector

You're all set. Choose your MCP client and follow the setup instructions.

Connector linkhttps://edge.vinkius.com/vk_preview_U02S7R70GaoTP6dztOIIMDtmgDyuN2GTH3Or66r0/mcp

Claude Desktop

Follow the steps below to connect in seconds.

  1. 1In Claude Desktop, open Settings → Connectors.
  2. 2Click “Add custom connector” and paste the connector link above as the remote MCP server URL.
  3. 3Click Add and start a new chat — Modelbit capabilities are ready to use.
Configuration · claude_desktop_config.jsonCopy
{
  "mcpServers": {
    "modelbit-ml-model-deployments-mcp": {
      "url": "https://edge.vinkius.com/vk_preview_U02S7R70GaoTP6dztOIIMDtmgDyuN2GTH3Or66r0/mcp"
    }
  }
}
Copy into chat04

  • Run inference on this text using my sentiment analysis model: [insert text]

  • Use my image classification model to check this data: [insert data]

  • Predict the next value in this sequence using my forecasting model: [insert sequence]

  • Check this customer feedback against my custom model and tell me the result.

  • Claude
  • ChatGPT
  • Cursor
  • VS Code
  • Windsurf
  • Claude Code
  • JetBrains
  • Cline

Start here

Connect Modelbit once, then ask.

One connection gives your agent access to all your deployed models. Your credentials stay encrypted and secure on our side.

Connect Modelbit to your AI

FAQ

How this task behaves

  • 01

    Can the AI deploy new models to Modelbit?

    No. The AI can only run inference on models that you have already deployed to Modelbit.

  • 02

    Does the AI store my model data?

    No. The AI only uses the data to perform the inference requested in the moment.

  • 03

    Can the AI change my model's parameters?

    No. The agent can only send data for prediction; it cannot modify the model configuration or weights.

  • 04

    What happens if the model returns an error?

    The agent will report the error from the model to you so you can troubleshoot the input or the endpoint.

  • 05

    Can I use any model from Modelbit?

    Yes, as long as the model is deployed and accessible via the Modelbit endpoint.

  • More questions about Modelbit? The Connector page answers them. See everything the Modelbit Connector can do

Connect Modelbit to Claude, Cursor, ChatGPT & more