Search...
ctrl/
Light
Dark
System
Sign in

JavaScript API​

@gel/ai is a wrapper around the AI extension in Gel.

npm
yarn
pnpm
bun
Copy
$ 
npm install @gel/ai
Copy
$ 
yarn add @gel/ai
Copy
$ 
pnpm add @gel/ai
Copy
$ 
bun add @gel/ai




Overview​


The AI package is built on top of the regular Gel client objects.


Example:


Copy
import { createClient } from "gel";
import { createRAGClient } from "@gel/ai";


const client = createClient();

const gpt4Ai = createRAGClient(client, {
  model: "gpt-4-turbo-preview",
});

const astronomyAi = gpt4Ai.withContext({
  query: "Astronomy"
});

console.log(
  await astronomyAi.queryRag("What color is the sky on Mars?")
);






Factory functions​



function
createRAGClient()
createRAGClient(client: Client, options: Partial<RAGOptions> = {}): RAGClient


Creates an instance of RAGClient with the specified client and options.




Arguments


    

  • 

    client – A Gel client instance.
    

    • 

  • 

    options.model (string) – Required. Specifies the AI model to use. This could be a version of GPT
                                            or any other model supported by Gel AI.
    

    • 

  • 

    options.prompt – Optional. Defines the input prompt for the AI model. The prompt can be
                                            a simple string, an ID referencing a stored prompt, or a custom prompt
                                            structure that includes roles and content for more complex
                                            interactions. The default is the built-in system prompt.
    

    • 













Core classes​



class
RAGClient


Instances of RAGClient offer methods for client configuration and utilizing RAG.




Ivar client


An instance of Gel client.









method
withConfig()
withConfig(options: Partial<RAGOptions>): RAGClient


Returns a new RAGClient instance with updated configuration options.




Arguments


    

  • 

    options.model (string) – Required. Specifies the AI model to use. This could be a version of GPT
                                            or any other model supported by Gel AI.
    

    • 

  • 

    options.prompt – Optional. Defines the input prompt for the AI model. The prompt can be
                                            a simple string, an ID referencing a stored prompt, or a custom prompt
                                            structure that includes roles and content for more complex
                                            interactions. The default is the built-in system prompt.
    

    • 










method
withContext()
withContext(context: Partial<QueryContext>): RAGClient


Returns a new RAGClient instance with an updated query context.




Arguments


    

  • 

    context.query (string) – Required. Specifies an expression to determine the relevant objects and
                                            index to serve as context for text generation. You may set this to any
                                            expression that produces a set of objects, even if it is not a
                                            standalone query.
    

    • 

  • 

    context.variables (string) – Optional. Variable settings required for the context query.
    

    • 

  • 

    context.globals (string) – Optional. Variable settings required for the context query.
    

    • 

  • 

    context.max_object_count (number) – Optional. A maximum number of objects to return from the context query.
    

    • 










method
async queryRag()
async queryRag(message: string, context: QueryContext = this.context): Promise<string>


Sends a query with context to the configured AI model and returns the
                        response as a string.




Arguments


    

  • 

    message (string) – Required. The message to be sent to the text generation provider's API.
    

    • 

  • 

    context.query (string) – Required. Specifies an expression to determine the relevant objects and
                                            index to serve as context for text generation. You may set this to any
                                            expression that produces a set of objects, even if it is not a
                                            standalone query.
    

    • 

  • 

    context.variables (string) – Optional. Variable settings required for the context query.
    

    • 

  • 

    context.globals (string) – Optional. Variable settings required for the context query.
    

    • 

  • 

    context.max_object_count (number) – Optional. A maximum number of objects to return from the context query.
    

    • 










method
async streamRag()
async streamRag(message: string, context: QueryContext = this.context): AsyncIterable<StreamingMessage> & PromiseLike<Response>


Can be used in two ways:


    

  • 

    as an async iterator - if you want to process streaming data in
    

    real-time as it arrives, ideal for handling long-running streams.
    

    

    • 

  • 

    as a Promise that resolves to a full Response object - you have
    

    complete control over how you want to handle the stream, this might be
                                            useful when you want to manipulate the raw stream or parse it in a custom way.
    

    

    • 





Arguments


    

  • 

    message (string) – Required. The message to be sent to the text generation provider's API.
    

    • 

  • 

    context.query (string) – Required. Specifies an expression to determine the relevant objects and
                                            index to serve as context for text generation. You may set this to any
                                            expression that produces a set of objects, even if it is not a
                                            standalone query.
    

    • 

  • 

    context.variables (string) – Optional. Variable settings required for the context query.
    

    • 

  • 

    context.globals (string) – Optional. Variable settings required for the context query.
    

    • 

  • 

    context.max_object_count (number) – Optional. A maximum number of objects to return from the context query.
    

    • 










method
async generateEmbeddings()
async generateEmbeddings(inputs: string[], model: string): Promise<number[]>


Generates embeddings for the array of strings.




Arguments


    

  • 

    inputs (string[]) – Required. Strings array to generate embeddings for.
    

    • 

  • 

    model (string) – Required. Specifies the AI model to use.
    

    •