[FREE PLUGIN] Release Google AI Gemini

[nocoderz]

Dear Bubblers,

I am happy to share with you that I have released a plugin for the new multimodal language Gemini!

Here is the link to the plugin page

Have a look at it and try it on your app
Demo page
Editor page

Let me know if you have any feedback

Eliott from nocoderz

[danschlung]

Hey Elliot. Thanks for building this plugin! Wondering if you have plans to add an action to the API that would allow us to post prompts from a workflow. (I’m having Gemini send me a JSON response that I’d like to pass into a backend workflow for data handling). I’ll likely just build my own API for now, but could be helpful to others if this were enabled, and/or you enable forking on the plugin

Anyway - really appreciate the work put into this & excited to put it to use in other parts of my application!

[nocoderz]

Hi Dansschlung,
Thank you for your feedback, I have just added actions for the existing data calls.
Let me know if it is good on your side!

[rohan2]

Hi Elliot,

Great work on the plugin.

I was hoping your plugin would help me run the Gemini Pro Vision…image and text.

I’m struggling to pass the image into the call.

Is your plug in able to help?

This is the code:

#!/bin/bash

Run this from the directory in which you saved your input images

This script was written to run on Debian Linux. It may require updates

to run on other platforms

Test that the first image is present

if [ ! -f “image0.jpeg” ]; then
echo “Could not find images in the current directory.” >&2
exit 1
fi

API_KEY=“YOUR_API_KEY”

curl
-X POST https://generativelanguage.googleapis.com/v1beta/models/gemini-pro-vision:generateContent?key=${API_KEY}
-H ‘Content-Type: application/json’
-d @<(echo ‘{
“contents”: [
{
“parts”: [
{
“text”: "As an expert in reservations, guest relations and hotel concierge, you know the best activities and tours to recommend to guests based on their passport photo. \n\nYour task is to extract the full name, nationality, gender, passport number and date of birth from the passport. \n\nYou will provide this information in a clear format required for hotel check-in. \n\nYou will then generate a list of recommendations for local tours, activities, sites and restaurants based on the guest’'‘s profile taken from the passport information. \n\nYou will be provided with:\n\nHotel Location: \nHotel Name: \nPassport photo of the guest\n\nYou should provide website links in your recommendations. \n\nYour tone should be quirky. Ensure one recommendation is a little risky and will make them laugh. \n\nHotel Location: Rawai Phuket\nHotel Name: Beach Boutique\n"
},
{
“inlineData”: {
“mimeType”: “image/jpeg”,
“data”: "’$(base64 -w0 image0.jpeg)‘"
}
}
]
}
],
“generationConfig”: {
“temperature”: 0.4,
“topK”: 32,
“topP”: 1,
“maxOutputTokens”: 4096,
“stopSequences”:
},
“safetySettings”: [
{
“category”: “HARM_CATEGORY_HARASSMENT”,
“threshold”: “BLOCK_MEDIUM_AND_ABOVE”
},
{
“category”: “HARM_CATEGORY_HATE_SPEECH”,
“threshold”: “BLOCK_MEDIUM_AND_ABOVE”
},
{
“category”: “HARM_CATEGORY_SEXUALLY_EXPLICIT”,
“threshold”: “BLOCK_MEDIUM_AND_ABOVE”
},
{
“category”: “HARM_CATEGORY_DANGEROUS_CONTENT”,
“threshold”: “BLOCK_MEDIUM_AND_ABOVE”
}
]
}’)

[nocoderz]

Hi @rohan2,
If I am not mistaken, it seems that you want to send an image encoded in base64 with a prompt. It is something possible with this plugin.
Please have a look at the following workflow (picture).

I hope this can help you implementing Gemini AI

[rohan2]

Elliot,

I really appreciate the time you are spending to help me.

This helps but I’m not sure how to pass it into the API call with Gemini.

I can use the Bubble API connector to integrate with Gemini Text only prompts but I’m not yet able to understand the whole flow to connect both.

Step by step would be great if you had the time. I just don’t have your skills I’m afraid.

Kind regards,

Rohan

[proteuscrypto]

Hello, greetings, I have been testing the plugin but I have had some problems with the response. For some reason I always get the same response no matter what image I send. My prompt is very basic. I only ask that you describe the image. I leave you here some images of the workflow and the responses. Thank you very much in advance for the response