DEV Community

Memo Chou
Memo Chou

Posted on • Updated on

GPT AI Assistant: Chat with AI using LINE App

About

GPT AI Assistant is a lightweight and extensible application that is implemented using the OpenAI API and LINE Messaging API.

Through the installation process, you can start to chat with your own AI assistant using the LINE mobile app.

Demo

GPT AI Assistant

Repository

https://github.com/memochou1993/gpt-ai-assistant

Installation

  • Log in to the OpenAI website.
  • Log in to the LINE website.
    • Add a provider (e.g. "My Provider").
    • Create a channel (e.g. "My AI Assistant") of type Messaging API.
    • Click the "Messaging API" tab and generate a channel access token.
  • Log in to the GitHub website.
    • Go to the gpt-ai-assistant project.
    • Click the "Star" button to support this project and the developer.
    • Click the "Fork" button to copy the source code to your own repository.
  • Log in to the Vercel website.
    • Click the "Create a New Project" button to create a new project.
    • Click the "Import" button to import the gpt-ai-assistant project.
    • Click the "Environment Variables" tab and add the following environment variables with their corresponding values:
    • OPENAI_API_KEY with the OpenAI API key.
    • LINE_CHANNEL_ACCESS_TOKEN with the LINE channel access token.
    • LINE_CHANNEL_SECRET with the LINE channel secret.
    • APP_LANG with en.
    • Click the "Deploy" button and wait for the deployment to complete.
    • Go to the dashboard, copy the application URL, e.g. "https://gpt-ai-assistant.vercel.app/".
  • Go back to the LINE website.
    • Go to the page of "My AI Assistant", click the "Messaging API" tab, set the "Webhook URL" to application URL with "/webhook" path, e.g. "https://gpt-ai-assistant.vercel.app/webhook" and click the "Update" button.
    • Click the "Verify" button to verify the webhook call is successful.
    • Enable the "Use webhook" feature.
    • Disable the "Auto-reply messages" feature.
    • Disable the "Greeting messages" feature.
    • Scan the QR code using the LINE mobile app to add as a friend.
  • Start chatting with your own AI assistant!

Upgrade

On your own gpt-ai-assistant project page, you can click on the "Sync fork" menu and then click on either the "Update branch" or "Discard commit" button to synchronize the latest code to your repository.

When the Vercel bot detects a change in the code, it will automatically redeploy.

Commands

Send commands using the LINE mobile app to perform specific functions.

General Commands

Name Alias Description
Talk /talk Talk with AI Assistant.
Draw /draw Ask AI Assistant to draw a picture.
Continue /continue Ask AI Assistant to continue the conversation.
Activate /activate Activate auto-reply. The VERCEL_ACCESS_TOKEN environment variable is required.
Deactivate /deactivate Deactivate auto-reply. The VERCEL_ACCESS_TOKEN environment variable is required.

System Commands

Name Alias Description
Command /command Show the application commands.
Version /version Show the application version.
Documentation /doc Show the documentation.
Report /report Report issues.
Restart /restart Deploy the application. The VERCEL_DEPLOY_HOOK_URL environment variable is required.

Sum Commands

Name Alias Description
Sum /sum Ask AI Assistant to give a "summarize" response
Advise /advise Ask AI Assistant to give a "advise" response
Apologize /apologize Ask AI Assistant to give a "apologize" response
Blame /blame Ask AI Assistant to give a "blame" response
Comfort /comfort Ask AI Assistant to give a "comfort" response
Complain /complain Ask AI Assistant to give a "complain" response
Laugh /laugh Ask AI Assistant to give a "laugh" response
Encourage /encourage Ask AI Assistant to give a "encourage" response

Analyze Commands

Name Alias Description
Analyze /analyze Ask AI Assistant to analyze
Analyze literarily /analyze-literarily Ask AI Assistant to analyze literarily
Analyze mathematically /analyze-mathematically Ask AI Assistant to analyze mathematically
Analyze numerologically /analyze-numerologically Ask AI Assistant to analyze numerologically
Analyze philosophically /analyze-philosophically Ask AI Assistant to analyze philosophically
Analyze psychologically /analyze-psychologically Ask AI Assistant to analyze psychologically

Translate Commands

Name Alias Description
Translate to English /translate-to-en Ask AI Assistant to translate text to English
Translate to Japanese /translate-to-ja Ask AI Assistant to translate text to Japanese

Environment Variables

Set environment variables to change program settings.

Name Default Value Description
APP_DEBUG false Print prompt to console. The value must be true of false.
APP_WEBHOOK_PATH /webhook Custom webhook URL path of application.
APP_LANG zh Application language. The value must be one of zh, en or ja.
APP_MAX_GROUPS 1 Maximum groups. The VERCEL_ACCESS_TOKEN environment variable is required.
APP_MAX_USERS 5 Maximum users. The VERCEL_ACCESS_TOKEN environment variable is required.
BOT_NAME AI Name of AI Assistant. This is used to call AI Assistant when auto-reply is deactivated.
VERCEL_PROJECT_NAME gpt-ai-assistant Custom Vercel project name. You can use this environment variable when the Vercel project name differs from the GitHub project name.
VERCEL_ACCESS_TOKEN null Vercel access token
VERCEL_DEPLOY_HOOK_URL null Vercel deploy hook URL
OPENAI_API_KEY null OpenAI API key
OPENAI_COMPLETION_MODEL text-davinci-003 Refer to model parameter for details.
OPENAI_COMPLETION_TEMPERATURE 0.9 Refer to temperature parameter for details.
OPENAI_COMPLETION_MAX_TOKENS 160 Refer to max_tokens parameter for details.
OPENAI_COMPLETION_FREQUENCY_PENALTY 0 Refer to frequency_penalty parameter for details.
OPENAI_COMPLETION_PRESENCE_PENALTY 0.6 Refer to presence_penalty parameter for details.
OPENAI_IMAGE_GENERATION_SIZE 256x256 Refer to size parameter for details.
LINE_CHANNEL_ACCESS_TOKEN null LINE channel access token
LINE_CHANNEL_SECRET null LINE channel secret

Click the "Redeploy" button to redeploy if there are any changes.

Debug

  1. Check if the environment variables of the project are filled out correctly in the Vercel.
  2. Click the "Redeploy" button to redeploy if there are any changes.
  3. If there is still a problem, please go to Issues page, describe your problem and attach a screenshot.

Related Projects

Top comments (1)

Collapse
 
jhunlimited_llc profile image
J H Unlimited L.L.C.

Well done mate. I have deployed to customers in Taiwan. Looking forward to collaboration in the future.