# LLMs could improve search with query-specific filters

Markdown copy of https://realalpha.app/s/llms-could-improve-search-with-query-specific-filters/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

- Speaker: [Aravind Srinivas](https://realalpha.app/builders/aravind-srinivas/) · Aravind Srinivas is cofounder and CEO of Perplexity. · Held research roles at OpenAI and Google DeepMind before co-founding the AI search engine Perplexity in 2022.
- Host: David Lieb · General Partner, YC
- Show: Lightcone
- Episode: [How To Build The Future: Aravind Srinivas](https://realalpha.app/v/SP7Ua8FKZN4/), Feb 21, 2025
- Moment: 3:43 to 6:42 of the episode (3 min), playing on https://realalpha.app/s/llms-could-improve-search-with-query-specific-filters/; the episode from there: https://realalpha.app/v/SP7Ua8FKZN4/?t=223

## Quote

> “And he was talking about how **LLMs could automatically figure out these suffixes**. And I was pretty interested in that.”
> — [Aravind Srinivas](https://realalpha.app/builders/aravind-srinivas/) at [4:44](https://realalpha.app/v/SP7Ua8FKZN4/?t=284) · Lightcone

## Takeaways

1. Add domain-specific suffixes to improve generic search rankings.
2. Use query classification to choose the right source.
3. Direct answers could reduce clicks that support search ads.

## About this segment

A blog on query reformulation gave Aravind Srinivas a way to improve Google results with domain-specific suffixes. He saw LLMs automating that step and later recognized the challenge to ad-driven search.

From “How To Build The Future: Aravind Srinivas”, Feb 21, 2025. [Watch the full episode](https://realalpha.app/v/SP7Ua8FKZN4/)

## Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[3:49](https://realalpha.app/v/SP7Ua8FKZN4/?t=229) How did you get the kind of activation energy to quit your great job at OpenAI and go do that? How did you find your co-founders? I came across this blog that one of the... Former YC partners Daniel Gross wrote, but it was like, how to build the next Google. And I think basically the core idea is like, you could do so much more with better query reformulation. So you take a query and you just add some suffixes.

[4:16](https://realalpha.app/v/SP7Ua8FKZN4/?t=256) So if someone's looking for reviews of a movie, just suffix side colon rottentomatoes.com. If someone's looking for reviews of some new gadget, do side colon that corresponding subreddit. You can get away with a lot of these suffixes and like our special strings to like filter results and already make Google so much better, even with the existing Google ranking. I'm not even talking about the ads problem, just simple ranking.

[4:45](https://realalpha.app/v/SP7Ua8FKZN4/?t=285) And then you can do more sophisticated things of like classifying queries. And he was talking about how LLMs could automatically figure out these suffixes. And I was pretty interested in that. Okay, that seemed like, okay, maybe generative AI might be, as in, like, I used to just call it LLM, so generative models could be a better way to build search engines, too. I also was pretty interested in, like, trying to do something agent-like when DeepMind had this Android environment that they built where, like, they kind of wanted to prototype a mobile app using agent that knows when to use what apps and control the apps.

[5:22](https://realalpha.app/v/SP7Ua8FKZN4/?t=322) that's when I spoke to my co-founder and CTO his name is Dennis we had written the same paper a day apart so we knew each other and he was a visiting student in my lab and we used to talk about I bring some ideas of how like we can build agents to control the Android environment so we were definitely chatting about doing a lot of things but never concretely about any company or product the first thing that anyone would tell you is like why why even work on this of course Google's going to do this right it's not even like you go build a better google docs uh and google will eventually do it because it's a secondary thing for them so companies like notion can still be funded this is their core crown jewel so why would you even try i think the reason it actually made sense this is again after launching the product we realized this not before um so there's like

[6:12](https://realalpha.app/v/SP7Ua8FKZN4/?t=372) some benefit to being ignorant, ignorance is bliss, is that if people stop clicking on links, the ad economy kind of dies. Now you can, there's a lot of nuance to this, but that core insight was only realized by us after launching. So once we realized that, I thought, okay, we were onto something and that kind of took us last two years. Walk us through like the first iterations of your experimentation, like I know you did