Aify

Thunderbird addon to bring ChatGPT to your emails.

Rewrite or generate text based on promps using OpenAI's API. Configure on settings page (make sure to save) and start to use when writing your emails!

You must have API key for OpenAI (anyone can access GPT-3.5-turbo). But then you can use any model you have access to including finetuned ones.

Aify comes with plenty of stock prompts, which you can add to, delete or customize as you need.

Screenshots

Make sure to configure the addon before use. You must at least set an API key and save it.

You can use Aify to turn sentences into detailed emails.

Or even translate and understand complex text.

Configuration and Usage

Configuration

1. Set an API key below (obtainable from OpenAI).
2. (Optional) Once you've set an API key you can click Update Models to get the current list.
3. Make sure you have access to the selected model (anyone can use gpt-3.5-turbo).
4. (Optional) Add, remove or edit the available actions.
5. Make sure to click save button to store settings.

Usage

1. Open a new compose window (Write, Reply or Forward).
2. Highlight some text to act one.
3. Click the Aify icon in the top right (near the lightbulb) and pick an action.
4. A popup will open and await the response from the API. This may take time at periods of high load.
5. (Optional) Click Regenerate to get a new response.

Tips and Tricks

You can add, edit and remove prompts from the settings page.

gpt-3.5-turbo is faster and cheaper than gpt-4, but not as "smart".

The prompt name (text box) is what is shown in the Aify context menu, and the prompt itself (the textarea) is prepended to the highlighted text separated by newline --- newline which makes it easy for the model to see whats what but isn't harmful if the prompt is empty.

Building

The contents of plugin/ should be zipped into a file ending .xpi.

cd plugin
zip -r ../aify.xpi *

Todo

• [HIGH] Use the API to identify available models.
• [MOD] You should be able to specify a model name (useful for finetunings).
• [HIGH] Limit length.
• [MOD] Model selection should be on a per-prompt basis.
• [LOW] Model selection should be selectable from the context menu.
• [MOD] Make it so any API could be used.
• [MOD] Styling and UX improvements.
• [LOW] Mask API key in settings once saved.

Privacy and Data Retention

This add-on interfaces with OpenAI's API but collects no data itself other than settings stored locally. The text is sent to OpenAI's server and managed by their data retention policy.

• OpenAI's privacy policy can be found at https://openai.com/policies/privacy-policy
• OpenAI's API data usage policy is here https://openai.com/policies/api-data-usage-policies

License

GPLv3

Icons

UXWing.com

Licensing terms - https://uxwing.com/license

• loading - https://uxwing.com/loader-line-icon
• icon - https://uxwing.com/invention-icon
• caution.png - https://uxwing.com/caution-icon