How SQREEM's Large Behavioral Model Predicts Consumer Intent with René Raiss

Summary

Marketecture: Get Smart. Fast.

Episode: How SQREEM's Large Behavioral Model Predicts Consumer Intent with René Raiss

Host: Ari Paparo
Guest: René Raiss, Founder of SQREEM
Date: June 30, 2026

---

Main Theme

This episode explores SQREEM’s unique approach to predicting consumer intent through its “Large Behavioral Model” (LBM), an alternative to language-based AI models. René Raiss discusses how SQREEM analyzes vast behavioral data without focusing on text or keywords, offering new ways to predict, segment, and act on consumer intent in advertising, healthcare, government, and beyond.

---

Key Discussion Points and Insights

1. What is SQREEM and the Meaning Behind the Name

• SQREEM (pronounced "Scream") stands for Sequential Quantum Reduction and Encryption Model.
• The name is a nod to quantum computing, though as René notes, “it was a good idea 15 years ago and we stuck with it.” (02:10)

2. The Core of SQREEM: Large Behavioral Model (LBM)

• Focus: Unlike Large Language Models (LLMs), SQREEM’s LBM centers on what people do, not what they say. It’s agnostic to language or keywords.
• Quote: “We don’t focus whatsoever on the written word or text. We are focused solely here on what people do, not what they say.” – René Raiss (02:44)

3. Understanding Behavioral Modeling

• SQREEM defines behavior as complex sequences of actions over time, which can infer intent, sentiment, and state of mind.
• Memorable Example: “One boring drink. Eight, nine different columns. Intent or, you know, adjacencies and behavior…You can’t capture that with a keyword, and it’s different for different people.” – René Raiss on the many meanings behind green tea (05:03)

4. Data Aggregation and Model Inputs

• Data-Agnostic Approach: SQREEM integrates data from 36 walled gardens (major social networks), web traffic from ~5 billion sites, open-source healthcare data, conferences, etc.
• Key Insight: The true value is in the model, not the uniqueness of the data.
• Quote: “You’re not going to get away in the future. It’s the model.” – René Raiss (06:57)
• Approach: Both top-down aggregation and bottom-up actionable identity mapping.

5. Product Offering and Use Cases

• Shift from Platform to AI Library: Rather than traditional platforms, SQREEM offers an AI library with functions producing specific outcomes (e.g., unmet customer needs, identity graphs, automated workflows).
• APIs & Integration: Brands can use SQREEM's APIs to generate predictive scores and intent wrappers for their first-party data.
• Use Cases:
  • Ad targeting and optimization
  • Market research and unmet needs analysis
  • E-commerce (over 5,000 campaigns, 91% lowered CPA) (09:17)
  • Biotech, financial services, and government policy modeling
  • “We help governments understand very often disenfranchised or hard to measure segments of the population…” (09:43)

6. Differentiators and Market Position

• Competitive Advantage:
  • “We mass produce intent, and we’re able to manifest that intent in any data spine or identity graph that exists in market today on demand.” (11:25)
• Biggest Challenge:
  • Misconception that success is due to special data, rather than modeling capabilities.
  • “The magic is what insights you can get out of a standard data set…the depth of that when it comes to information entropy and complexity of behaviors in the world is very deep.” (12:00)

7. Evolving User Experience

• The user experience has shifted; no more heavy platforms.
• Using endpoints, users state a problem and receive automated insights or dashboards generated “dynamically in query.” (10:28)

---

Notable Quotes & Memorable Moments

• On Green Tea and Human Behavior:
  “Green tea could be somebody’s journey towards flirting with veganism or vegetarianism...It could be an experience in a Japanese restaurant. It could be somebody’s concern about, you know, a health issue…One boring drink. Eight, nine different columns. Intent or, you know, adjacencies and behavior.” – René Raiss (04:50–05:11)

• On Data vs. Models:
  “The secret sauce is not the data...You’re not going to get away in the future. It’s the model.” – René Raiss (06:54–06:57)

• Impact in E-Commerce:
  “We have worked with a number of ecom players…we’ve done over 5,000 campaigns in E Commerce where the model was used today. And 91% of the time we lowered CPA.” – René Raiss (09:17)

• Competitive Advantage:
  “We mass produce intent, and we’re able to manifest that intent in any data spine or identity graph that exists in market today on demand.” – René Raiss (11:25)

• Animal Spirit of SQREEM:
  “Give me an animal where if the animal wins, nobody loses. So, a plant eater.” – René Raiss
  “Maybe like a nitrogen fixing bacteria.” – Ari Paparo (12:31–12:42)

---

Timestamps for Key Segments

• Company Name Origin: 02:00–02:24
• What SQREEM Does / LBM Concept: 02:37–03:16
• Deep Dive: Modeling Behavior: 03:27–05:11
• Model Inputs & Data Sources: 05:38–06:54
• From Platform to AI Library: 07:13–07:57
• Integrations & Predictive APIs: 08:07–08:48
• E-Commerce Results: 09:05–09:23
• Non-Ad/Media Use Cases: 09:35–10:06
• User Experience Evolution: 10:28–10:53
• Competitive Advantage: 11:19–11:36
• Biggest Challenge: 11:48–12:24
• “If SQREEM was an animal…”: 12:24–12:45

---

Conclusion

René Raiss gave an in-depth look at how SQREEM’s LBM harnesses massive behavioral data sets to predict real-world consumer intent and generate actionable insights for brands, governments, and researchers. The company’s shift from platforms to AI libraries, its modeling-focused approach, and its impressive track record in campaign performance suggest a compelling new direction in behavioral targeting and market analytics.

Memorable closing exchange:
Ari Paparo: “That’s the first time we’ve had a bacteria as the animal of the interview.” (12:45)

---

For more, visit Marketecture.tv.

Transcript

[A]

Oh, hello, I'm Jeremy Bloom, co founder and CEO of marketexture Media. And boy do I have news for you. Market is back. And if you've been watching from a distance, thinking that looks phenomenal, but a trip to New York is just a bridge or several too far. We've got great news for you. We're bringing Market live to the beating heart of Adland. That's right. On September 23, 2026, market coming to Chicago. We're going to be bringing the same sold out energy, sharp insights and industry defining conversations to the center of advertising's biggest transformations. From media and commerce to AI tech and modern marketing. This is where the people shaping what's next for the ad industry will be, want to be and need to be. Early registration is live now, so lock in your [e-mail] Again, lock in your ticket at chicago.architecture live.com disclaimer. I live in Chicago. It's not Chicago, it's Chicago.

[B]

Welcome to Mark and Tech show where you can get smart fast with in depth interviews of leading technology executives. I'm Ari Paparo. I am recording from the Cannes UTA beach and my guest today is Renee Reis who is the founder of Scream. Rene, thank you so much for being here.

[C]

Ari, thank you very much for having me. It's very, very exciting to be in Cannes. It's even more exciting to finally get to meet you in person.

[B]

Yeah. Well, it is nice to meet you. I appreciate you also have a good voice radio like you're. It's. You sound good.

[C]

Now you're making me self conscious.

[B]

Okay, so we have to start with the obvious thing which is how do you spell the name of this company? Because it's Scream, but it's with a Q. So take me through the name.

[C]

Scream Sqr eem. Sequential quantum reduction and encryption model. It's a bit of a contradiction if you really want to go into quantum physics, but you know, it's. It was a good idea 15 years ago and we stuck with it.

[B]

Okay, well, we think of this podcast as being a bit technical, but not that technical. We're going to stay outside of the quantum realm and deal with Newtonian physics as it relates to advertising. So what does Scream do?

[C]

So we started this a while back and unapologetically it's taken more than a decade. We built a foundational model, but really the core of what we're focused on is behavior. Okay, so we call it a large behavioral model because we don't focus whatsoever on the written word or text. We are focused solely here on what people do, not what they say.

[B]

That's interesting. So it's LLM. One of the L's is language, right? Yes. So that is not what you do.

[C]

Correct. So we, we go with lbm.

[B]

LBM is that real large behavioral model. Okay, so what do you mean by behavior? Are we talking about visits to websites and things like that?

[A]

That's.

[C]

That's a very good question. And, you know, I. I promise I'll. I'll not go there, but if you think of behavior for a second in physics, it's a complex system moving forward in time or many complex systems.

[B]

Okay.

[C]

And behavior is activity, activity that leads to actions, actions that come in sequences over time and eventually form behaviors, different behaviors for different people, different sequences. And you could take it one step further and say that that becomes intent, that becomes sentiment, that becomes state of mind, and it branches from there.

[B]

So if you, if you could track, you know, everything I did throughout my entire life, which you'd obviously, just for sake of a mental argument here, you might be able to better predict the next things I might do, that you would know that I like coffee and that I tend to have a coffee at like one in the afternoon. Hey, it's one in the afternoon. Probably high probability.

[C]

You know, interesting. There is that even though you and I may have a coffee at one in the afternoon, you are in my behavioral trip to get there, or behavioral journey, so to say, is going to be unique to us.

[B]

Right?

[C]

So the idea that people try to wrap keywords and buckets and folders around here is intent for life insurance. And therefore everybody in this bucket is intending to buy life insurance because they've gone to this website. It's not how the real world works. And that's what we're trying to solve.

[B]

Well, how does the real world work? What insights are you learning?

[C]

You know, behavior is not reliably contextual. If you think of something as simple as green tea, it's boring. Green tea.

[B]

Green tea.

[C]

Green tea could be somebody's journey towards flirting with veganism or vegetarianism. It could be somebody wanting to lose weight.

[B]

Right.

[C]

Somebody wanting to go to, you know, Ubud Bali to a spa on an Eat, Pray, Love getaway. And green tea is part of that. It could be an experience in a Japanese restaurant. It could be somebody's concern about, you know, a health issue, prostate cancer, antioxidants.

[B]

Right.

[C]

One boring drink. Eight, nine different column. Intent or, you know, adjacencies and behavior.

[B]

Right.

[C]

You can't capture that with a keyword and it's different for different people.

[B]

So, so your model is now going to the CMO of the green tea company and saying we could predict on various, various vectors of green tea drinking what might be, might be appropriate for a consumer.

[C]

Yeah, a single consumer. It could be different moments in time, it could be different segments that are evolving through time. But yes, in essence, green tea is the most boring one we could think of.

[B]

I don't think that's kind of an interesting one. So can you give me a little insight behind the curtain as to what the behaviors that make up the model are? Are they like, you know, web activities, purchase activities, outdoors, you know, just like in a big, big picture.

[C]

Yeah. So it's basically if you, if you think there's a, there's a top down aggregated layer and then there's a bottom up layer because at some point these behaviors need to become actionable and become some form of identity where. Or a moment that allows you to target the individual.

[B]

Right.

[C]

But if you start at the top, there is no. We're data agnostic. So what that means is we basically aim to take as much as possible. So we've got integrations with 36 walled gardens, all the big social networks. We have a technology that allows us to infer aggregated traffic data to websites. So we're doing that at large scale, about 5 billion websites. We do a bit of web crawling to contextualize traffic.

[B]

Right.

[C]

Maybe just to, I mean to, to organize it and then beyond that in different markets. If we're going into healthcare, you know, there's open source data around, you know, drug interactions, you know, patient experiences, you've got professional associations, conferences, congresses. But the key area to all of this is that the secret sauce is not the data.

[B]

Right.

[C]

You're not going to get away in the future. It's the model.

[A]

It's.

[C]

You have a set of data that's probably large but non unique.

[B]

It's fascinating from a perspective of a customer or potential customer. What am I getting? Am I just licensing model and saying like, hey, I need to use this for predicting visits to my stores or something like that.

[C]

So we're at a very, very interesting junction in the world where if we had this conversation a year ago, I would be using the word platform 40 times in this conversation. Sure, everyone loves a platform and we're done with platforms. Not a platform, not a platform. Think of us as a, an AI library. So there are functions and these functions deliver outcomes. It may be an outcome as specified as what are unmet needs for customers, enter brand.

[B]

Right. Okay.

[C]

It could be something as generic build me, you know, an identity graph for people who are intending abc.

[B]

Okay.

[C]

So there are outcome functions around insights, there are functions that, you know, automate a workflow and there are functions that produce an outcome in a third party platform like NADSP or Social Graph or things like that.

[B]

But it can be used for boring use cases like targeting ads to people.

[C]

Yes, we're boring people.

[B]

Boring is the wrong word. Easy to understand use cases. Let's go that way. So in that case, let's say I have a bunch of first party data about my consumers. Do I Basically use your APIs to create predictive scores about what they're about to do or what their behavioral likelihoods.

[C]

Correct. So you have first party data and most of the time we'll call this static or stationary.

[B]

Okay.

[C]

Unless you're very, very, very real time E commerce.

[B]

Okay, sure.

[C]

What we in that scenario become is an intent wrapper. You may have all the data of travel. We'll tell you when Mr. ABC or Mrs. ABC are intending to fly to Miami on a weekend golf getaway, that intent is where we come in. On top of that deterministic data, have

[B]

you measured, say how impactful that is versus other methods?

[C]

Yeah. Data about whether or tremendous amount? Yes.

[B]

You want to share any?

[C]

Well, I mean it, it work depends on, you know, the outcome that somebody's trying to build on top of the model. But we, we've run this in 80 countries around the world.

[B]

Oh really?

[C]

We've run this across health care, B2B consumer, some very, very niche area, some very, very generic. We have worked with a number of ecom players where the closed loop on conversion is very simple.

[B]

Right.

[C]

I think We've done over 5,000 campaigns in E Commerce where the model was used today. And 91% of the time we lowered CPA.

[B]

That's amazing. Right?

[C]

It's hard work. It's taken a decade. This is, we didn't come up with this.

[B]

You're the, you're the original found?

[C]

Yes.

[B]

All right, so what are. So you've mentioned a bunch. Are those the most common use cases?

[C]

We have use cases where the model is used emerging right now in biotech, far away from media, financial services.

[B]

Yeah.

[C]

We help governments understand very often disenfranchised or hard to measure segments of the population. You know, in Europe, South America, Australia, things like that. So there are use cases where it is used AI for good.

[B]

Okay.

[C]

There are use cases where it's used in very, very deep research on the Consulting side, biotechnology, medical. But our biggest vertical, of course, is media.

[B]

So a government use case might be like, if we pass this policy, how are our citizens going to, you know, use it or something along those lines.

[C]

Along those lines, yes.

[B]

Right. That's fascinating. It's the same model, but it's the same model, Correct? Right.

[C]

It is behavior, but governments struggle as much as brands do. Understanding the world.

[B]

How. How advanced or technical does a customer need to be to use your product?

[C]

So again, this is a very good question that in the last six months, the answer has changed dramatically. We're now in the world of mcps. In other words, you have an endpoint where you've asked a question or stated a problem, and we have a response where an MCP server will basically find a number of functions in Scream and solve your problem and generate generatively a dashboard for you. So all of this happens dynamically in Query. There's no more platforms, there's no more training.

[B]

Right, fascinating. So let's move to kind of a lightning round. I'll just ask you some quick questions. You give me some quick answer. So first is, what are you hoping to accomplish at cat?

[C]

Well, this is a big step for us because we have started to scale quite aggressively into market. We've been in stealth mode. We've been in a number of markets outside the US Testing this for the last few years. But we are coming into the world now very aggressively.

[B]

Yeah. Exciting. And what is your number one competitive advantage?

[C]

We mass produce intent, and we're able to manifest that intent in any data spine or identity graph that exists in market today on demand.

[B]

That's really. It's a little hard to get your head around it. Right. Because I'm kind of technical and still. It's like I'm still kind of trying to. Trying to mold it. All right, what is your biggest challenge?

[C]

The biggest challenge is that the market still has the idea that if. If I'm making a bold statement to you, Ari, most people say, oh, you must have access to data that nobody else has.

[B]

That's a good question. Yeah.

[C]

Tell me your secret data. Well, so I can verify it. The magic is what insights you can get out of a standard data set. And the depth of that when it comes to information entropy and the complexity of behaviors in the world is very deep. So it's the model. To your point earlier, it is not the data, but we have an obscene amount of data that comes with that. It's just none of it is special.

[B]

Fascinating. Final question. If Scream was a animal, what animal would it be?

[C]

Give me an animal where if the animal wins, nobody loses. So a plant eater.

[B]

A plant eater, maybe. Or maybe like a nitrogen fixing bacteria.

[C]

Yeah, something. Yeah, that's, that's really good.

[B]

Okay. That's the first time we've had a bacteria as the animal of the interview. All right. Those fascinating. So, Renee Reis, the founder of Scream, Scream spelled S Q, R, E, E M. Thank you so much for joining me in. Can you.

[C]

It's been amazing. And thank you so much for having me, sir.