Microsoft.Extensions.AI Integration

Cross-SDK comparison

See the centralized MEAI documentation for feature matrices and comparisons across all tryAGI SDKs.

The Predibase SDK provides AIFunction tool wrappers compatible with Microsoft.Extensions.AI. These tools can be used with any IChatClient to give AI models access to Predibase's LLM deployment management, LoRA fine-tuning job creation and monitoring, and dataset browsing.

Installation

dotnet add package Predibase

Available Tools

Method	Tool Name	Description
AsListDeploymentsTool()	ListDeployments	List all private serverless LLM deployments with status and endpoints
AsGetDeploymentTool()	GetDeployment	Get details of a specific deployment by name
AsCreateFinetuningJobTool()	CreateFinetuningJob	Create a new LoRA fine-tuning job with base model, dataset, and config
AsGetFinetuningJobTool()	GetFinetuningJob	Get fine-tuning job status by ID
AsListFinetuningJobsTool()	ListFinetuningJobs	List all fine-tuning jobs with statuses and timestamps
AsListDatasetsTool()	ListDatasets	List all uploaded datasets with row counts and column info

Usage

using Predibase;
using Microsoft.Extensions.AI;

var predibaseClient = new PredibaseClient(
    apiKey: Environment.GetEnvironmentVariable("PREDIBASE_API_KEY")!);

var options = new ChatOptions
{
    Tools =
    [
        predibaseClient.AsListDeploymentsTool(),
        predibaseClient.AsGetDeploymentTool(),
        predibaseClient.AsListFinetuningJobsTool(),
        predibaseClient.AsListDatasetsTool(),
    ],
};

IChatClient chatClient = /* your chat client */;

var messages = new List<ChatMessage>
{
    new(ChatRole.User, "Show me all active deployments and their fine-tuning job statuses."),
};

while (true)
{
    var response = await chatClient.GetResponseAsync(messages, options);
    messages.AddRange(response.ToChatMessages());

    if (response.FinishReason == ChatFinishReason.ToolCalls)
    {
        var results = await response.CallToolsAsync(options);
        messages.AddRange(results);
        continue;
    }

    Console.WriteLine(response.Text);
    break;
}