Let’s get a little bit technical with AI today
Since I plan on using Firebase for my backend, I thought, why not also use Gemini to power my app? Maybe it will make my life simpler if all my backend is powered by Google (even though we know OpenAI has the more powerful model)
Google also just recently announced a bunch of AI products in their Google I/O, so this might be a good time
I would also be able to submit an entry to the Google AI Competition alongside buildspace (talk about hitting many birds with one stone, haha)
So, I went into the Google AI rabbit hole, and oh boy was I overwhelmed!
Here are their products:
- Firebase ML: A suite of ML tools for mobile and web app developers
- Gemma: A family of open models used to develop the Gemini models
- Gemini: Their flagship large language model (LLM)
- Google AI Studio: A web-based platform for building and deploying ML models without coding
- Google AI Edge: A suite of tools for running ML models on edge devices
- Firebase Genkit: A tool for generating code for Firebase projects
- Gemini in Firebase: A service that helps Firebase developers build, launch, and optimize high-quality apps
- Vertex AI: A unified machine learning platform for building, deploying, and managing ML models
Confused yet? you’re not alone, haha
Maybe they should tone down the use of the terms AI and ML in their product naming
Anyway, I will have to work with RAG (Retrieval-Augmented Generation), this is a fancy term for “using your own data with AI”
This is different from “fine-tuning,” which is a fancy term for converting a general AI to a specialized AI (which would require a lot of training and experimentation with data)
With this, I hope Apple would also release an On-device Gen AI SDK so I don’t have to setup an AI backend infrastructure and can focus more on the app’s functionality
This, however, will introduce another problem of feature disparity across platforms, but that’s another problem for another day once Apple has had their say
Tomorrow, let’s get less technical and have more fun
Leave a Reply