Wow, thank you so much for the feature! Incredibly honoured to have my words shared by you all.

Some of us were given the opportunity and taught to turn tools and workflows into code and applications that actually help others. This is very empowering and reveals a new opportunity to grow, honouring those who taught us by becoming teachers and empowering others ourselves 🙌.

chatGPT did disappoint but let's see if Google now can surprise with a new Gemini model the amount of areas they utilise AI in is incredible. Very curious!

Great work looking forward for more! 🙌🥳

I try to think how this must've been when the computer just came out. People would've likely called themselves computer specialists. That's us now! 🥳😂

Just in case you are curious I actually wrote an article about this particular demo use case: www.scheppening.com/article/2/gi...

Yeah I think browser engine and screen sizes it can potentially already consider if its connected to a playwright browser MCP Server. Therefore it starts headless browsers (can do multiple browser engines) to look at a locally run web application visually and compare with e.g. a figma layout.

I agree as soon as it's something AI does not have a reference video or any context for, it completely relies on descriptions in the prompt and does a lot of guessing work that tend to lead it towards wrong directions quite often.

Really nice to hear and gives a lot of confidence!

For me as AI enthusiast it makes it easier to share context when I hand it an URL. Appreciate that Google needs to make sure it needs to respect robots.txt agent configuration to prevent it's AIs from reading content when it's not allowed to which is probably what you are referring to is it?

Very interesting read. Thanks for sharing!

But the 80 or 90% it does relatively fast. I take it for skeletons and the rough build it's doing quite a good job you just gotta experience what you already did knowing it needs a manual look over the code to roughen out any more detailed bits. That fair?

So happy to hear! Thanks for the answer and can't wait! 🙌

For completion: OpenAI as well as Google won gold with their models

Would be great if you could connect to a created custom model trained with a Vertex AI corpus. Just realized you are actually working on gemini-cli at Google. Cool to find you at bluesky. Can't wait for more! 🙌

What are the results so far?

Does it not have one context in the pro version without having to use this command when quiting and restarting Gemini cli?