---
title: "Search the Web as It Was: An Introduction to Exa Snapshot"
speakers:
  - name: Ryan Ahern
    affiliation: Exa
room: "Shannon Room (NY Ballroom E)"
starts_at: "2026-10-13T11:30:00-04:00"
ends_at: "2026-10-13T11:48:00-04:00"
format: "talk"
recording: "2026-10-13_1130_Exa"
duration: "18m 53s"
source: "qwen3-asr-1.7b"
---

# Search the Web as It Was: An Introduction to Exa Snapshot

> Transcribed by qwen3-asr-1.7b with a schedule prompt. Possible prompt artifacts, names heard more often here than in the unprompted Voice Memos transcript (Qwen vs Voice Memos): Exa (8 vs 0). Check them against the audio before quoting.

### 00:00:00 — From Human Keyword Search to Agentic Web Search

And today I'm going to give you an overview of two things. We only have a very short amount of time. So the first one, I just want to motivate what is Exa? What is Agentic Web Search? Just because you kind of need to believe in the dream of that, understand why you want to use it to the second thing I'm going to talk about, which is how to backtest financial models using that.

So first, just to understand where I'm at in the room, who here has heard of Exa before or used it? Okay, like a good show of hands. Who here has ever used like Google back in the day to like do some sort of work or personal use was like typing into Google to do some sort of search? Hopefully, like everyone in the room. People had pretty good market share. You might have heard of it. Last one would be, how many of you now no longer really use Google? You use ChatGPT or Claude or maybe even Instinct or Muse if you're like at the cutting edge to actually do your searches? Yeah, most people, right? So we're going to connect all of these things together.

This last point, where a lot of people now like, hey, I don't even really use desktop search anymore, I kind of ask an agent to do this, is something that we see more broadly in the market, and the idea is that like agents tend to do a pretty good job at understanding how to search because they're allowed to try multiple times. They kind of don't get tired. They tend to be quite agreeable and work very hard. So when you say like this is the thing that I want, it's going to go out and it's going to do many different searches in order to satisfy like your objective, and the other thing here is that like they tend to search different things.

So as a human, you might put in like a few keywords for the thing, and then hope that you get the best results, like in the first couple, and otherwise like maybe you try like a different querying, then you kind of give up. Those agents will be very diverse. They tend to describe what they want to search in great detail, execute many searches, and want to understand the space more fully before they pass, and that means that they do way more searches on a post basis. So generally, the idea is like, hey, nowadays most humans, you're going to be using agents to and broker your search tasks, and that's where ExaTrader plugs in. So when you're working with like many agentic systems or whatever cursor, Logsinger and such latest documentation to help us creating agents for this project, whenever you're like talking to Gamma, you're like, hey, it'd be great if I had just this like SEC data from like Apple in like 2018 to complete my slide, like that would make it brilliant. It's going to use next step to dirt out such the full web, so we have our crawling infrastructure, our database, our embedding model, our retrieval architecture to get exactly what it needs to answer its query.

This is like a couple of logos of like the bigger companies, but because this is AI engineer, like here, a few like kind of archetypes that we work across the financial. I work with a lot of hedge funds, asset managers, sovereign wealth funds, P Cs, across like a bunch of different tasks because they also need like the latest information. So if you are a hedge funder, you're going to be trading on information. You're looking in the market for alpha, right, and for information that allows you to understand how you could position your organization better. If you're a an insurer, you're probably understanding what other parties you're working with, or who you're insuring, to make sure that ultimately, you're making the determination, so you want to make sure you're making the best information and that your risk models are the most accurate. We're even here in VC. I need the hottest startup, I need to know the index instinct before they blow up, and they're not necessarily going to be like some disrupted dataset, like I need to go and find them wherever they exist in the web, and that's what they come to Exa to kind of support them for.

### 00:03:31 — Three Core Tenets: Quality, Dynamic Highlights & Low Latency

And then the kind of to wrap up the last part of my answer and the reason why I think now you'd be like, okay, I'm building agents. I'm interested in agents, and I definitely need an agent website for my website. This speech is like we're a lab in the sense that we're very interested in the search market, and in the ways in which agents search, they're hyper focused on that problem. There's three like core tenants, and then we'll talk about how that goes into versioning to segue into the next part of the talk, but first and obviously is that quality, right? So you want to make sure that when the agent does that the tool call and the retrieval; it gets what it needs because the fewer tool calls that it needs to make, the more cost effective on your inference, and also just like the more likely is to succeed on its overall trajectory. You're not going to bring something that misaligns it.

I'm really, really interested in all these guys. We recently released, like a new benchmark, which we think more accurately assesses the performance of like the search tools separately, so they can because these two things can be combined: like what agent or what model is pulling the search, and what search is going into what model. So that you can kind of have a look at that to see how they're thinking there. Secondly, on cost, so it's not just the line item of the tool call, right? Like how much does one search cost? Because agents perform many different searches.

So the idea is both like, hey, what's the value statistical thing of trajectory, but also, like, what's the impact of this on the total cost of that agent completing its task? Because that's actually the things that you care about, and you generally get opinionated about one tool call. It's like, how does this overall agent take? And that's why, like we intro, we work on things like, say, dynamic highlights, which means that for every single result, it may not be not just returning full pages, we're not even just taking one page, which is our previous highlight features, and figuring out what are the exact right snippets on that page that we need to send back to the context window to complete the trajectory, we now do that across all of the search results, which means that you can actually get higher quality results at ninety-five percent fewer tokens than you would at full-page text. We have a blog about this. It's very interesting. It basically means you'll save a lot of money downstream on all of the input tokens you'll be spending on inference for the agents to complete its trajectory.

Last one on latency. So because we own our infrastructure, we can do very low latency search. I'm just going to touch on the other side of the spectrum, which is like very long-running deep research. So we have Exa Agent Ultra, which we release and like not too long ago, maybe a few weeks ago, which is like state of the art on a lot of benchmarks around like very long running search, and that's because we can do a lot of things under the hood. We have obviously we bring the search kind of so we can also be opinionated about that, how that interacts with the agentic partners to do this like very long running research task. So I highly recommend you check those things out. This is just a segue to be like, okay, Exa really cares about this space, I think that their search is going to be good. So moving to the next part, which is really interesting, is the topic of this talk, which is versioning.

### 00:06:29 — Introducing Exa Snapshot: 400 Billion Versioned Web Pages

So just recently, we released Exa Snapshot. It's 400 billion plus snapshots of the web, and you can search them by day. And that's really interesting in the in the financial services industry. We'll get into that, but first, I'm just going to talk a little bit about like what this basically means. So we're a search company, so we're constantly searching across the web for the latest information that we can add to our index. And every single time we do it, we check it for we're checking for freshness because ultimately you want to have the latest live up-to-date information at the point at which you're searching, but because we're doing all of those checks, it means that we can capture a version of those URLs at the point where we checked it and have like a version history, and you can access that in two ways. One is like I know the URL, I want to find as it was or before a given date.

And so you can give us a URL, and you say this is the date that I want it. I want it the affidavit before and serve it to me as you would have if I had requested it at that point. The second one, which is even more exciting, is this concept of version stuff. So where such API. And the idea is that hey, what would happen if I had executed that search? That has very like interesting implications. I'm just going to do one like very small example here.

### 00:07:50 — Point-in-Time Revenue Signals & Avoiding Leakage in Backtests

So, OpenAI — recently, let's say as of October first. On the left-hand side here, you see a snapshot search as it was in October first, and on the right-hand side is a live search as of like a couple of days ago when I made this slide, but at that time, in early October, like there were some discussions that like OpenAI had a like annualized like run rate around like the seventy million, right? And that was leading to certain ways in which the market was projecting how and where they would be valued, and then more recently there was some, and that was quite revolutionary back that year.

And then more recently, there were some publications that were reporting like actually it's closer to fifty billion for financial accounting reasons, right? Like it turns out that they were counting gross revenue when actually, like some of their revenue was coming from cloud providers and technically, the way that you should account for that would be that there would be a member pass throughs and net annualized revenue is better. That's how WhatsApp what some other players are reporting, etc., etc. But basically meant that the net annualised run rate was kind of down, forecast from seventy billion down to fifty billion. Right now, this is just like a financial thing that some of you may or may not find interesting, but the so what version such as like, let's say you're in the interest of creating a product or a solution or an agent that is made to that will that that's a job is to find these things, right?

You can now go back and be like, okay, before everyone figured it out, right, before the market and before some financial analysts published the think piece that was like slamming down on this particular thing, and everyone knew about it, what information existed on the headset, right? Like, was there an SEC filing that would have revealed this thing in an accounting footnote? Like was there some other reporting from some other vendor that maybe is like you would consider as like less reputable, but actually, was able to find that kind of thing, and maybe you should consider them differently. You can potentially like trace these things before like the rest of the market found it.

But most importantly, like not only can you test that, but you could backtest that, right? You could then adjust the ways in which you are setting up your model. You can adjust the ways that you're setting up your agent to be able to be could have found this discrepancy, and more importantly, couldn't have found that discrepancy in a way that is generalizable, right? Let's like create a set of like fifty sample sets, see if it can perform against that. Check it against the holdout. Be like, okay, I think that my model is now going to be pretty good at finding this kind of feature, and that's something that's kind of new in the in the in the web search space and that you can backtest that. This is like a like a little diagram just to kind of motivate the idea.

But at any like live search that you do with like any existing search product, the idea is it's live, so it has all of the information. So in this particular little example, we're trying to predict whether a company got acquired or not. In this case, if you do a live search, you're going to find like that the deal was announced, right? Because that happened. It got released everywhere. You could be like, hey, I'm trying to train a model that's able to predict whether this company got acquired or not. So I'm gonna go like one day before the announcement, seven days before the announcement, a month before the announcement, and see whether I could or could not predict it, right? And you can build a set of those companies and basically train your model or whatever system it is that's trying to do this prediction, and see like, was it able to reliably I could predict this particular feature.

### 00:11:18 — Three Backtesting Archetypes: Enriched ML, Agentic Features & Full Agents

So I'm going to spend a little bit of time here just talking about how different systems are currently being backtested with Snapshot. So there's like three kind of archetypes here. The first one is I am operating like a very traditional model — maybe it's ML. Maybe it's even more simple than that, like a decision tree or something. And all I want to do is just gather unstructured data from the version web into some sort of structured set of information. So it could be something like software prices. It could be some sentiment analysis. It could be like the terms and conditions of a particular site.

Whatever it is that like powers your model, not all information is continually available. Like if you're trying to understand software prices, like you know, the page of that software rights not necessarily stored in a version a way for that website in a way that you could reliably access. Or potentially, like you have some, like you know, sentiment analysis that you want to do, but like that's going to be mired in like what point was that sentiment released in time, right, and you want to be able to get like a reliable set of those signals, and then you can just transform it into some array signals that goes into your model, you can predict it. That's like the sort of basic space, and some people are just like enriching data that they wouldn't have had access to for their existing models. So that's just like sort of like the status quo plus one. The interesting space is in like the next two areas.

So the first one is this idea of like adding an agentic feature, and the I and here is like okay well I have this existing model, it performs pretty well, but I have an idea that like if I add one or two like agentic features to this model, it would be able to perform better. So let's say something like an adverse media scan. Necessarily, like adverse media is going out into the web, searching across it for all the information that exists, potentially doing it in an agentic way, right, where you're doing multiple searches, you're exhausting multiple different trajectories, and then you're going to take all of that and then condense it into some sort of structure.

Another example here would be to say, like equity research, in much the same way, like, hey, I want to go out, I want to look at the full universe of companies, and I want to find like potentially all of their like estimated, like rack revenue or headcount or all of the features that matter, and the idea is that you'd be using something like ExaSearch at the point at which you're doing a prediction like to go out and gather all of that information to then send into your traditional model, and you can now use like extra snapshot to backtest those particular features, so you can go back and be like, hey, if I was to do this like adversarial scan, like one month before, or maybe I have already like onboarded a bunch of people into my portfolio, like four months ago, what would I have said at the point where they were going to enter my portfolio, and how has that tracked against today?

And that kind of motivates, like especially with financial services customers, like you don't now just have to believe that like search will be better. You can actually go back and back test that against your existing processes and say, hey, it was better. So that's like maybe the next evolution. It's like one step plus.

On the far right-hand side, here is kind of like maybe the future. So there's some like startups that are working in this production space, and the idea here is that the full sort of like simulation or prediction is being managed by an agent. So in the middle is like an ML model that's maybe doing a prediction or some algorithm or decision tree. On the right hand side, like it is like an agent. This is going off and doing a bunch of things. It could be searching internal knowledge. It could be accessing data feeds. It could be doing whatever it is, but one of the things that it may do that is valuable is to talk of and access the web, and the idea is like now you don't have to try and create some internal corpus that tries to represent what it would be like to access. Then you can just be like, hey, I want to make sure that I can control that feature so I know that future information is not just going to leak directly into the model. Like I'm trying to predict company acquisitions through some very insane, like agency processes, accepting everything.

You can now be like, all right, let's just make sure that's not just gonna have one web search that finds that it was compliant, comes back and predicts it with like 100 percent accuracy. So that's kind of the sort of right hand side there.

### 00:15:36 — Workshop Acquisition Signals, Overfitting & Complex Semantic Features

I'm just gonna give like a little example of like the middle, just to round off this kind of discussion. So yesterday we ran that a kind of extended workshop on this kind of shot, and we had a bunch of around 100 people who were that were kind of running through this same example, we had before, trying to predict company acquisitions. One thing that was kind of interesting is that this was like one of the example features in the workshop, and one of the people that that's kind of relevant. So the thing that was interesting here is like, yeah, that just to motivate. So the first one was this idea that like hey, within the past year, there was a investor or a significant shareholder publicly for personal sale company as a whole, and the second one, which is the interesting one, it's like, hey, did this company share price was a recent set by the highest ever pass by the S D I S, it kind of dips down, and the last, the interesting thing is like, is the company name? Now this is a little bit of overfitting. People that know like backtesting, like and the idea here is like this could be something like a patient could do right, just based on like the lifestyle, and potentially just overfit to the training data.

The idea with snapshot is like not only can you go back and test for these things, you can go back and like set up your emails and your agency contacts in such a way that you can detect these things, but you've also got to check that it generalizes. And this is something that will be very hard to do give you unvalidated by like testing in production. And you can only change the ways in which these features are described over time, and check against real time, right? You might be waiting for months to try and see how and why your agent set up actually is predictive or is overfitting in different ways.

Then he's like, you can describe these like quite complex features, like these ones are like maybe you could imagine trying to recreate in like some sort of like non-sentient such, but you can go as crazy as you want within existing like Exasemantic search world. So you can say things like, Hey, I want to see if there was any adverse media against like reputable VPs that are operating, like that are operating within VC funds that have like been on the AI Engineer podcast, and they say something bad about it, right? Like you can very as hard as you want on these features that you think are going to be predictive, because that's what core existential allows, and then you can access it and see its performance.

### 00:18:04 — Recap: Agentic Web Search & Point-in-Time Backtesting

Why not? Then I'm just gonna kind of like recap. So the first thing that we talked about here is like using Exa because it's quality plus effective agentic web search across every latency. We have lots of blogs around there to encourage you to not try next, ah, to have a go at this, because the first thing you want to know is like, hey, do I actually think this is going to have some benefits for my model? The second things we talked about is like, hey, using Exa Snapshot to be able to like backtest these financial models, or general model, risk model, prediction model with these web search features in order to see, like, hey, does adding it add value to you and your agentic processes? Thanks much.