Docs chat

Intellinode cloud allows you to connect your data to various chatbot engines, including OpenAI ChatGPT, Google Gemini, and LLama V2. This integration enables a tailored chatbot agent experience, providing tuned responses to the context of your uploaded documents or diagram images.

How setupt intellinode cloud with your data

1. Visit the IntelliNode App.
2. Start a project using the Document option.
3. Upload your documents or images, such as PDF, DOC, DOCX, PNG, JPG, and code files.
4. Copy the generated One Key; this key will be used to connect IntelliNode's chatbot to your uploaded data.

Implementation of Chatbot Models

The setup code is identical for all bots, making it easy to switch between them. You can employ your One Key with different language models as shown in the examples below:

First, import the necessary modules:

const { Chatbot, SupportedChatModels } = require("intellinode");
const intelliKey = '<generated_one_key>';

Assuming that your data set includes software contracts, you can ask the chatbots details about the contract using this code:

let query = "List to me the included features in the vector database contract";

OpenAI ChatGPT

Incorporate the One Key with chatGPT in the following way:

const openaiBot = new Chatbot(openaiKey, SupportedChatModels.OPENAI, null, {oneKey: intelliKey});

You can ask the ChatGPT bot details about the data using this code:

const { ChatGPTInput } = require("intellinode");

const input = new ChatGPTInput();
input.addUserMessage(query);

const responses = await openaiBot.chat(input);
responses.forEach(response => console.log("- " + response));

Google Gemini

To configure Google Gemini, use the One Key in this way:

const geminiBot = new Chatbot(geminiApiKey, SupportedChatModels.GEMINI, null, {oneKey: intelliKey});

You can interact with the Gemini bot using this code:

const { GeminiInput } = require("intellinode");

const input = new GeminiInput();
input.addUserMessage(query);

const responses = await geminiBot.chat(input);
responses.forEach(response => console.log("- " + response));

Mistral AI

To configure the Mistral open source model with your data, use the One Key:

const mistralBot = new Chatbot(mistralApiKey, SupportedChatModels.MISTRAL, null, {oneKey: intelliKey});

You can interact with the Mistral bot using this code:

const { MistralInput } = require("intellinode");

const input = new MistralInput();
input.addUserMessage(query);

const responses = await mistralBot.chat(input);
responses.forEach(response => console.log("- " + response));

LLama V2 - Replicate

To implement LLama V2 - Replicate Bot with the One Key, use:

const replicateBot = new Chatbot(replicateApiKey, SupportedChatModels.REPLICATE, null, {oneKey: intelliKey});

Interact with the LLama V2 bot with the following code:

const { LLamaReplicateInput } = require("intellinode");

const input = new LLamaReplicateInput("You are a helpful assistant!");
input.addUserMessage(query);

const responses = await replicateBot.chat(input);
responses.forEach(response => console.log("- " + response));

Advanced Options

Improve your chatbot responses with additional customization through optional parameters:

• searchK: specifies the number of references for the semantic search step.
• attachReference: includes the names of reference documents with the chatbot's responses.

Example Incorporating Advanced Options with OpenAI ChatGPT

Here's how to utilize these advanced options in an OpenAI ChatGPT integration:

// initiate the chatbot
const openaiBot = new Chatbot(openaiKey, SupportedChatModels.OPENAI, null, {oneKey: intelliKey});

// setup the input with the advanced options
const input = new ChatGPTInput("you are helpful assistant", { searchK: 4, attachReference: true });
input.addUserMessage(query);

// when sending attachReference you should use result to get the content
const responses = await openaiBot.chat(input);
responses.result.forEach(response => console.log("- " + response));

// get the referenced documents
console.log('### the chatbot references ###')
console.log(Object.keys(responses.references))

Example Incorporating Advanced Options with Gemini

Here's how to utilize these advanced options in an OpenAI ChatGPT integration:

// initiate the chatbot
const geminiBot = new Chatbot(openaiKey, SupportedChatModels.GEMINI, null, {oneKey: intelliKey});

// setup the input with the advanced options
let query = "List to me the included features in the vector database contract";
const input = new GeminiInput("you are helpful assistant", { searchK: 4, attachReference: true });
input.addUserMessage(query);

// when sending attachReference you should use result to get the content
const responses = await geminiBot.chat(input);
responses.result.forEach(response => console.log("- " + response));

// get the referenced documents
console.log('### the chatbot references ###')
console.log(Object.keys(responses.references))

This example demonstrates how to activate document references (attachReference: true) and set the number of references (searchK: 5). This approach can be adapted to any supported chatbot model, enhancing the contextual understanding.