Valentin Sawadski

@vsaw
639 Followers
1,098 Following
1.3K Posts

Focusing on the future, it's where I'll spend the rest of my life!

- #Startup & #VC 🚀
- #IoT & #AI 🤖
- #Sustainability 🌍

Currently combining all of the above #CTO of https://getvamo.de, to help people make the switch to #heatpumps. #HeatYourHomeNotYourPlanet

Prev: Co-Founder https://tado.com & CTO https://proglove.com

Side Project: https://opendata.cam

Websitehttps://vsaw.net
GitHubhttps://github.com/vsaw
Dev.Tohttps://dev.to/_vsaw
Twitterhttps://twitter.com/_vsaw
Valentin Sawadski2d ago

Okay, Migration Assistant is interesting. I wonder though assuming I migrate a GPL licensed app from iOS to Android. As the new app will most likely be ported from Swift to Kotlin, so most likely not a single line of code made to Android, so would GPL copyleft still apply? Or could I release the app in any license (provided no other trademark is violated)

#GoogleIO

Valentin Sawadski2d ago

RE: https://mastodon.social/@vsaw/116603014428457501

Aha! The #GoogleIO developer demo shines some light on it, as they demo how to create a fine-tuned version of Gemma-4.

I assume they will use multiple fine tuned models for different use cases internally, potentially even on an "per agent" level?

Show threadValentin Sawadski2d ago

Does anyone know why Google went and built dedicated silicon for #AI model training? (Improving their Gemini models or an extended zoo of (personalised?) models?)

A bit of googling shows the original announcement is almost a month old now, but it does not go into detail what exactly the TPU 8t models will be doing 🤔

https://cloud.google.com/blog/products/compute/tpu-8t-and-tpu-8i-technical-deep-dive 2/2

#googleIO

TPU 8t and TPU 8i technical deep dive | Google Cloud Blog

The 8th generation TPUs are engineered with system-level co-design to accelerate the AI lifecycle. TPU 8t is built for frontier-model training and TPU 8i is built for large-scale inference and reinforcement learning.

Google Cloud Blog
Valentin Sawadski2d ago
The biggest news for me at #GoogleIO today was TPU 8 with their specialised version for Training and Inference. Making custom silicon is a serious effort, so Google must see a future whey they are constantly training models on a massive scale. I wonder if all this is to develop their next Generation of Gemini Models, of if we soon a much larger model zoo of (personalised) models? 1/n
Valentin Sawadski2d ago

Wait? That’s it? Nothing on #GoogleBook where they put Gemini right into the OS, or anything for Android besides the ominous „Android Halo“ mention?

#googleio

Valentin Sawadski2d ago

This whole Shopping Demos are a mess. First they introduce new shopping protocols, now they show a demo that does not use them but instead does AI automation for exiting Android apps. Feels like multiple teams competing with each other or a rushed announcement.

#googleio

Valentin Sawadski2d ago

So let me get this straight all we get as non-US people is a bigger search box?

Omni, Spark all other new features so far are behind the walled garden of the US border…

#googleio

Valentin Sawadski2d ago

I like how Google envisions and describes the Agent Experience.

- Personal
- Proactive
- Powerful

#googleio

Valentin Sawadski2d ago

Finally MCP support for Gemini!

#googleio

Valentin Sawadski2d ago

I wonder how the business model behind Universal Commerce Protocol, AP2 and Universal Cart looks like for Google

(The reason behind it is quite clear: Google fears that agentic shopping is eating up their ad revenues as it happens in the Agent interface and not the browser anymore)

#googleio