The first step in our selection rate optimisation pipeline is done. We can now identify the most valuable sentences in any piece of content on any page, at scale. The 'most valuable' means the most likely to be picked by Google to form grounding snippets for Gemini 3. I'm ecstatic about this and can't wait to take it further.
What’s the logic on how text is selected here Dan?
Super interesting work, Dan. This sparks a really important question for brands...what happens when the sentence Google sees as most valuable isn't the one that's the most accurate? Most brands aren't ready for that shift yet.
Well that’s incredibly valuable
Damn that is going to a molecular level 👀
Gemini 3 is truly a game changer :)
Test run please? 😊
👀👀
Nice! Are generating the synthetic prompts based on the provided URL’s content (or something else I could not see in the image)?