PaLM API: Chat Quickstart with Swift

Overview

This quickstart demonstrates how to use the PaLM API, which gives you access to Google's latest large language models, with the Swift SDK. You'll use the chat service of the PaLM API which is a model variation built for dialog-focused use cases such as chatbots.

Prerequisites

The latest version of Xcode running on MacOS.¹

Setup

Install the API client

The following instructions will guide you through downloading and installing the PaLM Swift SDK, which also includes the sample app you will be using in this tutorial.

git clone https://github.com/google/generative-ai-swift

Obtain an API key

Follow the instructions on the setup page to create an API key for your app. You will need this API key in the next step.

Try out the sample application (Optional)

Navigate into the Examples folder in your cloned generative-ai-swift directory and open the PaLMChat app in Xcode.

  cd generative-ai-swift/Examples/PaLMChat
  xed .

1. While in Xcode, press CMD+B to build the app. When building the app for the first time the build script will generate a new file called PaLM-Info.plist.

2. In Xcode, open PaLM-Info.plist and paste your API key into the API_KEY setting

   

3. Press CMD+R to run the app. The phone simulator will launch and show a simple chat UI. Enter a message into the input field and tap the Send button to start a conversation with the model.

Use the SDK in your app

Add the SDK package to your project

To use the PaLM API in your own app, add the GoogleGenerativeAI package to your app:

1. Create a new Swift app.
2. Right-click on your project in the project navigator.
3. Select Add Packages from the context menu.
4. In the Add Packages dialog, paste the package URL search bar: https://github.com/google/generative-ai-swift
5. Click on Add Package. Xcode will now add the GoogleGenerativeAI package to your project.

Initialize the API Client

Before you can make any API calls, you need to import and initialize the API client.

1. Import the GoogleGenerativeAI module:

   import GoogleGenerativeAI
   

2. Initialize the API client:

   let palmClient = GenerativeLanguage(apiKey: "YOUR API KEY")
   

Call the Chat service

Now you're ready to call the PaLM API's chat service. Call the chat method with a prompt to get a response from the model. See the Intro to LLMs guide to learn more about prompting.

let userMessage = "Tell me a joke"
let response = try await palmClient.chat(message: userMessage)
if let candidate = response.candidates?.first, let responseText = candidate.content {
  print(responseText)
}

Continue the conversation

Calling the chat method will return a GenerateMessageResponse, which contains a history of the current conversation between the user and the model as an array of Message objects.

To continue the conversation between the user and the model, take this array, append the latest response candidate from the model, and pass it back to the model for the next turn of the conversation.

if let historicMessages = response?.messages {
  history = historicMessages
  history.append(candidate)
}

// ... later
// send previous chat messages *and* the user's new message to the backend
response = try await palmClient?.chat(message: userMessage.message, history: history)