Author: admin

In a recent interview, Google’s VP of Product for Search, Robby Stein, shared new information about how query fan-out works in AI Mode.

Although the existence of query fan-out has been previously detailed in Google’s blog posts, Stein’s comments expand on its mechanics and offer examples that clarify how it works in practice.

Background On Query Fan-Out Technique

When a person types a question into Google’s AI Mode, the system uses a large language model to interpret the query and then “fan out” multiple related searches.

These searches are issued to Google’s infrastructure and may include topics the user never explicitly mentioned.

Stein said during the interview:

“If you’re asking a question like things to do in Nashville with a group, it may think of a bunch of questions like great restaurants, great bars, things to do if you have kids, and it’ll start Googling basically.”

He described the system as using Google Search as a backend tool, executing multiple queries and combining the results into a single response with links.

This functionality is active in AI Mode, Deep Search, and some AI Overview experiences.

Scale And Scope

Stein said AI-powered search experiences, including query fan-out, now serve approximately 1.5 billion users each month. This includes both text-based and multimodal input.

The underlying data sources include traditional web results as well as real-time systems like Google’s Shopping Graph, which updates 2 billion times per hour.

He referred to Google Search as “the largest AI product in the world.”

Deep Search Behavior

In cases where Google’s systems determine a query requires deeper reasoning, a feature called Deep Search may be triggered.

Deep Search can issue dozens or even hundreds of background queries and may take several minutes to complete.

Stein described using it to research home safes, a purchase he said involved unfamiliar factors like fire resistance ratings and insurance implications.

He explained:

“It spent, I don’t know, like a few minutes looking up information and it gave me this incredible response. Here are how the ratings would work and here are specific safes you can consider and here’s links and reviews to click on to dig deeper.”

AI Mode’s Use Of Internal Tools

Stein mentioned that AI Mode has access to internal Google tools, such as Google Finance and other structured data systems.

For example, a stock comparison query might involve identifying relevant companies, pulling current market data, and generating a chart.

Similar processes apply to shopping, restaurant recommendations, and other query types that rely on real-time information.

Stein stated:

“We’ve integrated most of the real-time information systems that are within Google… So it can make Google Finance calls, for instance, flight data… movie information… There’s 50 billion products in the shopping catalog… updated I think 2 billion times every hour or so. So all that information is able to be used by these models now.”

Technical Similarities To Google’s Patent

Stein described a process similar to a Google patent from December about “thematic search.”

The patent outlines a system that creates sub-queries based on inferred themes, groups results by topic, and generates summaries using a language model. Each theme can link to source pages, but summaries are compiled from multiple documents.

This approach differs from traditional search ranking by organizing content around inferred topics rather than specific keywords. While the patent doesn’t confirm implementation, it closely matches Stein’s description of how AI Mode functions.

Looking Ahead

With Google explaining how AI Mode generates its own searches, the boundaries of what counts as a “query” are starting to blur.

This creates challenges not just for optimization, but for attribution and measurement.

As search behavior becomes more fragmented and AI-driven, marketers may need to focus less on ranking for individual terms and more on being included in the broader context AI pulls from.

Listen to the full interview below:

---

Featured Image: Screenshot from youtube.com/@GoogleDevelopers, July 2025. 

In a presentation at the 2025 Kidney Cancer Research Summit, Thomas Powles, MBBS, MCRP, MD, spoke with CancerNetwork® about the results of a Quality-Adjusted Time Without Symptoms or Toxicity (Q-TWiST) analysis from the phase 3 LITESPARK-005 trial (NCT04195750) comparing belzutifan (Welireg) with everolimus (Afinitor) in patients who were heavily pretreated with clear cell renal cell carcinoma.¹

LITESPARK-005 was a randomized phase 3 study designed to evaluate belzutifan, a HIF-2 alpha inhibitor, against everolimus in this patient population. Although previous findings from LITESPARK-005 showed a high response rate and improved progression-free survival (PFS) with belzutifan, an overall survival advantage was not demonstrated. However, the Q-TWiST analysis showed that belzutifan exhibited a distinct toxicity profile and correlated with better overall quality of life.

The rationale behind conducting this Q-TWiST analysis was to gain a more nuanced understanding of patient benefit beyond traditional endpoints like PFS. According to Powles, this analysis is crucial because while PFS indicates the duration patients are progression-free, it does not account for the impact of treatment-related toxicities on their quality of life. By integrating quality-of-life questionnaires, Q-TWiST may provide a valuable perspective on the patient experience by focusing on the period without progression and without toxicity.

The findings from this analysis showed that belzutifan outperformed everolimus due to a longer time to progression and a longer time without toxicity, which was deemed clinically meaningful.

Powles is a professor of genitourinary oncology, lead for Solid Tumor Research, and director of Barts Cancer Institute at St. Bartholomew’s Hospital at Queen Mary University of London.

What was the background for conducting this Q-TWiST analysis among patients who received everolimus or belzutifan in the LITESPARK-005 trial?

LITESPARK-005 was a big, randomized phase 3 study. It tested belzutifan vs everolimus in patients who were heavily pretreated with advanced clear cell renal cancer. Belzutifan is a HIF-2 alpha inhibitor. It’s well tolerated. It’s an agent that is associated with anemia, some hypoxia, and fatigue. Compared with VEGF tyrosine kinase inhibitors [TKIs], in my clinical experience, it’s well tolerated. [The trial] showed high response and better PFS [with belzutifan]. We didn’t show an overall survival advantage. We showed a distinct toxicity profile, and we also showed better overall quality of life [with belzutifan].

In this Q-TWiST analysis we performed, what we were looking at was the time that the patients had without progression and without significant toxicity. That’s essentially what it is; PFS just looks at progression and the area of patients who are progression-free, which is great, but if all that time was spent with significant toxicity, that means less than if a patient has the same amount of time without [PFS] with no toxicity. We can construct Kaplan-Meier curves, and within those curves, we can define specific areas: those areas without progression and [with] toxicity, those areas without progression and without toxicity, and those areas after progression. In the Q-TWiST analysis, specifically, we look at that time, that period, and that area without progression and without toxicity, and then we bring in the quality-of-life questionnaires to work out what, in terms of that area, the quality of life is during that period. That’s the Q of Q-TWiST.

That’s a summary of how it works. Essentially, it’s not a surprise that belzutifan outperformed everolimus in this respect quite significantly. The time to progression was longer, the area was therefore larger, and the time without toxicity was longer. That was clinically meaningful by previous parameters, which had been set to perform Q-TWiST analysis.

What did findings show about the quality-adjusted survival time in patients who received belzutifan vs everolimus in this population?

The story behind belzutifan, so far, is an agent that has been developed in 2 areas that are contrasting with each other. No. 1 is Von Hippel-Lindau [VHL]–driven disease, and that disease has shown that renal tumors can be put into remission or regression associated with belzutifan, preventing surgery on primary renal tumors. That’s important for patients. In the other extreme, in metastatic clear cell renal cancer, we go all the way where it’s been developed very late: the third or fourth line, in many cases. Those 2 are very different from each other. The development of the disease—the VHL disease—and how that relates to patients who are very heavily pretreated is quite complicated. It’s probable, or even likely, that if we use belzutifan earlier in the disease process, it will work better. What we’ve shown with this Q-TWiST analysis is not only that it is more active than everolimus, but it appears to be, from a quality-of-life perspective and a time-dependent quality of life analysis, to be better tolerated, making it an attractive [option].

What are the next steps for researching belzutifan or other therapeutic agents among patients with previously treated advanced renal cell carcinoma?

The story of belzutifan is a long one, and looking forward into the future, there are several randomized trials [that are ongoing]. There’s the adjuvant [phase 3 LITESPARK-022 study (NCT05239728)], and as I said before, [belzutifan] might work better earlier in the disease.² Pembrolizumab [Keytruda] plus or minus belzutifan in patients who are high-risk post-surgery is No.1. It’s in the frontline setting….It’s also [being assessed] in a randomized phase 3 trial [NCT04586231] in patients who received 1 prior therapy, essentially second-line treatment, which is lenvatinib plus belzutifan vs cabozantinib [Cabometyx].³ [There is] a plethora of randomized trials earlier in the disease and big randomized studies. Belzutifan has a promising future, and because it’s well tolerated, that makes it an attractive combination.

What do you hope others take away from your presentation?

The main goal for them would be to say the analysis confirms our impression and our clinical experience that belzutifan is well tolerated. My personal experience is that it’s true. These data should not come as a surprise to anyone that our Q-TWiST analysis for belzutifan is better than everolimus, and I hope that they also see it as not just an attractive option in patients who are heavily pretreated, which is the case. If you switch away from VEGF TKIs to belzutifan, what you’ll find is that the toxicity profile—the chronic fatigue and diarrhea associated with VEGF TKIs—can be switched off by [choosing belzutifan]. It does have a role to play. I accept it doesn’t have an overall survival signal by itself. The response, the [PFS], the quality of life, and these Q-TWiST data give a compelling argument for this being an attractive agent in patients who are heavily pretreated.

References

1. Powles T, de Velasco G, Choueiri TK, et al. Quality-adjusted time without symptoms or toxicity (Q-TWiST) analysis of belzutifan versus everolimus in previously treated advanced renal cell carcinoma (RCC): LITESPARK-005 (LS-005). Presented at the 2025 Kidney Cancer Research Summit; July 17-18, 2025; Boston, MA. Abstract 13.
2. A study of belzutifan (MK-6482) plus pembrolizumab (MK-3475) versus placebo plus pembrolizumab in participants with clear cell renal cell carcinoma post nephrectomy (MK-6482-022). ClinicalTrials.gov. Updated July 17, 2025. Accessed July 22, 2025. https://tinyurl.com/y69wn2xb
3. A study of belzutifan (MK-6482) in combination with lenvatinib versus cabozantinib for treatment of renal cell carcinoma (MK-6482-011). ClinicalTrials.gov. Updated April 2, 2025. Accessed July 22, 2025. https://tinyurl.com/mtftnfpt

The Nintendo 64, which launched in 1996, had an unusual three-prong controller design, with a single analog joystick in the center of the gamepad above the middle grip. The 8BitDo 64 offers the same buttons as the N64’s controller in a more traditional layout, but while it’s easier to hold, the lack of a second joystick limits its compatibility with modern games, which generally expect two. The 8BitDo 64 is compatible with the Switch, Switch 2 (after updating the controller’s firmware), Windows, and Android devices and connects to consoles, PCs, and mobile devices using either a Bluetooth connection or a USB cable.

It also features more durable Hall effect joysticks and shoulder buttons which should help prevent it from developing stick drift over time. But if you’re not planning to use it with the Analogue 3D (someday), the controller will potentially only be useful for playing classic N64 games through an emulator or the Nintendo Switch Online service but your GoldenEye 007 muscle memory may not perfectly transfer over to the updated button layout. Nintendo’s own wireless N64 controller is a more accurate reproduction, but you still need to be a Nintendo Switch Online subscriber to buy one.