App: https://app.lingostand.com/

Summary

The main objective of this release was to integrate a Large Language Model to Lingostand. To accomplish this, I focused on a simple requested feature: Show the IPA phonetics per sentence. Even though I dont know IPA phonetics, I liked this feature becuse:

• It is a feature long requested by Fabian user’s
• Currently there are few places to get phonetics IPA for a sentence, tophonetics.com is the main one.
• It allowed me to compare 2 main LLMs: Palm 2 (used by Google Bard) VS GPT
• Undertand what was needed to integrate them
• Update our Python AI container to support the requirements
• Get inspiration for future ideas and ground the next UI conversational experiment.

I would say that the iteration was a success… For the next iteration I will use GPT 4 turbo, and the experimental feature Assistence. But I wrote several powerful and straightforward use cases for Palm2 model.

New Features

UI

• IPA phonetics per sentence using GPT

Minor improvements

Website

Updated the webpage to start preparing for the new features to be added. Since there was so much to remove my idea was to do a first iteration to remove things that we are striping away from the app, and estart experimenting with the upcoming offering.

The idea was also to prepare it to update the webpage regualry every other release or so.

UI

• “New Version” screen created to communicate to the returning users new features in version changes

Infrastructure

Python AI container

• Container was broken and unable to deploy. It was fixed by upgrading the Docker version
• Python upgrade form 37 to 39
• Python AI

Backend

Translation

• google translation brough back instead of deepl translator

Email Metrics Service to admins

• Metrics service simplified by deleting totals and excel attachment