Hi. I just started with GLIDE and I am looking to build an app and use BI, ML, AI. Anyone in the community with experience in such things. Please contact me if yes.
I’m interested in connecting. I have rebuilt an app in Glide that started as a standard dev project with standard dev tools (I just ran out of money and my co-founder had to step out).
The app I am working on was using AI and ML in a few ways:
- extract meaning from text - specifically we wanted to extract 5 - 10 keywords from text strings inputted by the users (not too hard to do)
- learn from user behavior to steer them to more relevant content (ML)
I am now playing around with Amazon Rekognition (built into Cloudinary) to see what fun there could be with auto-tagging images. And looking into what can be done when you have to drive things off a GSheet. Many things to explore!
I will DM - thanks - and welcome to the Glide community
Interesting initiative. The biggest problem here is how to make the data flow smooth and not affecting user experience, specifically loading time for your ML/AI model to process user’s input data.
I imagine you have to notify the model on any new data that needs to be processed, then let them process it, write it back to the Sheet and then wait for Glide to sync it back to the app.
Indeed - but it’s not necessary to have things happen live in almost all cases.
For instance there needs to be a card ‘Deck’ built every day. This would contain all possible cards any user could be swiping on, then use filters to personalize the experience.
Cards might be inputted automatically at the Gsheet backend (via RSS) and the model runs periodically for keyword extraction. Then you could run the personalization engine with ML overnight based on recent swipes, compared to the users existing model. None of this needs to be done real-time
Tks for the info. As told, I am new and starting. All tips are welcome.
Tks for support and info.
I am too interested in how Glide can be used to run Machine learning models. I am new to Machine learning but have been running scripts in python for machine learning models. When you mentioned you run the personalization engine with ML overnight, how do you connect glide to run the machine learning code automatically. I am not an expert in programming, but is the running of the machine learning model done at the google sheet end or through Glide itself using a webhook?
good question if find out let me know.
Good question. I am not an expert neigher. But I can suppose that probably we can write and use the script in the sheet used by Glide and via this script to connect it with ML or with another script linked to ML. Let’s developers correct me if I am wrong and for sure the feasibility.
Correct. For instance I have played with MonkeyLearn, ParallelDots and MeaningCloud in GSheets as add-ons. It looks like some are run manually, some can have custom formulas, and some can be run as scripts.
I could have a process overnight that extracts keywords from relevant text in each row and add it into a ‘Keywords’ column.
I could then run a custom script to integrate new keywords from users into their existing set (using ‘magic’) and create a revised keyword algorithm every morning.
And then run another script to decide what Cards to make available to each user.
I could even run an Amazon Rokognition algorithm on Cloudinary to extract tags from any uploaded photo… to have another source of keywords to play with.
Glide is not doing any of this. It is then the presentation and user interface layer
Great to know about this. It would be awesome if I could create a flight price prediction app using Glide as the presentation and user interface layer with the data mining running in the background. A whole new world of AI or Data science apps can be available if this can be connected to Glide. Keen to see how this develops up.
Here’s some exciting news for implementing AI models in Google Sheets. Check this video and peltarion.com.
This makes glideapps based AI a possibility.
Integrate Peltarion AI models in Google Sheets for sales forecasting