José Mauricio Duque: Before we begin, why don't you present the system that you've created?
Robert Caulk: Yeah, we're really excited to come out with this first set of publishers joining the new system. We have a vision for how the relationship between publishers and the AI world should interact.
It's a very difficult bridge to build. There are copyright protections, licensing agreements, monetization tracking, optimizing ingestion on the LLM side. All of these form one bridge, and that's what we call the AskNews bridge: bridging publishers to the AI ecosystem.
In our bridge, we're focused on protecting data using a synthetic representation. This grounded synthetic document takes the original text and extracts the most important context, while stripping the voice, narrative, and original expression. Instead, we build a contextual document, which we call grounded synthetic data, index it, make it searchable, and pass it to the AI ecosystem.
That's key to the AskNews bridge. It protects original publisher rights. We can dive deeper, but to us, this really distinguishes us from other news monetization platforms. They're great in their own way, but we're focused on creating that clean bridge.
José Mauricio Duque: That's great, Rob. What I see is that this system makes publishers visible. With other systems, the publisher becomes invisible, right?
Robert Caulk: It depends, yeah. Retrieving relevant information is key. AI ecosystems (these LLMs) speak in natural language and need a very specific subset of relevant articles at any given moment to reason on top of. That retrieval process, getting those grounded synthetic documents out of the database at the right time, puts publishers in front of the right consumer.
Other systems might not have an index. It's more crude or raw. Our approach is more like Spotify: a curation of information that the LLM needs at that moment. We curate instead of just giving access like the Cloudflare model, where users scrape publisher websites. That burden falls on the AI user, but we find they prefer curated, cleaned information. As long as the source is relevant, they want it delivered in a ready-to-use format.
As long as the source is relevant, they want it delivered in a ready-to-use format.
José Mauricio Duque: That makes sense. Would it be fair to say your system makes it easier to trace and attribute publisher content?
Robert Caulk: Yes. The publisher isn't necessarily attributing their own content, but when their content surfaces in our system for the AI side, the AI knows which publishers and articles are included. That context helps the LLM make smarter decisions.
Traceability is crucial. Even though the original text isn't passed along, a synthetic representation is. Someone down the AI pipeline is going to want to verify where the information came from. So yes, passing along traceability is key.
José Mauricio Duque: How does your collaboration with Newstex strengthen your ability to scale licensed news data, and what unique value does it bring compared to direct partnerships?
Robert Caulk: That's a good question. We get it a lot from publishers. Licensing is a laborious part of the process. With big publishing houses, deals can take six months with a lot of back and forth. That slows down our strength, which is engineering and research.
We're scientists. We focus on synthesizing information so AI can ingest it: how to store, retrieve, and scale it. Newstex's strength is interacting with publishers. They've been doing it for over 20 years, maintaining relationships, handling the long conversations around licenses.
Newstex's strength is interacting with publishers.
Offloading that to Newstex is valuable. They also have clients with diverse use cases. Each publisher can choose which clients get access to their data. We're just one additional market, serving B2B, high-stakes decision-making systems in AI. That's how this partnership lets us scale. We each focus on what we do best.
José Mauricio Duque: So Newstex handles the relationships and royalty payments.
Robert Caulk: Exactly.
José Mauricio Duque: Let's talk about the AI-ready news economy. AskNews serves research and high-stakes decision platforms that need factual, machine-readable signals from trusted news, not prose. How does your synthetic news layer differ from traditional aggregation, and why is it essential for enterprise AI?
Robert Caulk: That segment of the market wants a clean, contextual picture. The cleaner the picture, the better the LLM can reason and make decisions.
Dumping HTML into an LLM gives messy input: ellipses, links, irrelevant text. It's not optimal. Plus, tokens drive cost, so the more you input, the more expensive the decision.
Our synthetic layer takes HTML, processes it, enriches it, adds context, and outputs a high-resolution image of the world. Let's say an LLM wants info on EU trade tariffs. It gets clean, contextual content from multiple languages, countries, sources, industries.
That gives a high-resolution model of what's happening, enabling forecasts like whether tariffs will change. Our layer adds geolocation, entity extraction, relationships, publish dates. All the clues an LLM can reason on. It's objective, diverse, and compact. Better input equals better output. It's quality in, quality out.
Better input equals better output. It's quality in, quality out.
José Mauricio Duque: That's a great way to put it. And it's not only more accurate, it's cheaper, faster, more comprehensive.
Robert Caulk: Exactly. And more reliable. Businesses want predictable outputs: today, tomorrow, ten years from now. That's why sustainability matters. Scraping HTML isn't sustainable. It's brittle for engineering and brittle for journalism. Journalists need funding to continue. Licensing helps keep that model alive.
José Mauricio Duque: With your 50% revenue-sharing model, how are publishers responding compared to other AI licensing deals? And what usage patterns are you seeing through the Publisher Dashboard?
Robert Caulk: We're not alone in offering 50/50 revenue shares. Others like Perplexity do similar things. But the key difference is we don't pass original text. That matters to publishers.
Some models allow scraping, but we use synthetic data to protect originality. As for the dashboard, it gives publishers analytics on how and when their articles surface. They can see revenue, yes, but also insights like: "Why did one article appear 1,000 times and another zero?"
This helps inform editorial strategy. We've seen clustering around certain queries, forecasting markets like Polymarket, Kalshi, and Metaculus. These platforms track real-time events, and our system feeds them diverse, multilingual data. Topics range from product launches to geopolitical shifts. The best publishers are those reporting on major global events. That's what the system picks up on.
José Mauricio Duque: The global AI market is projected to hit $107 billion in 2025. What use cases in forecasting, risk assessment, and finance are driving the demand for structured news data?
Robert Caulk: News analytics are key to many industries testing systems in the AI market. It's still early, but we see strong use cases.
In risk assessment, clients are using our data to assess regional security, prepare reports, or issue tactical briefings. These are human-in-the-loop applications. Analysts aren't going away. They're being enhanced.
In finance, news APIs have always been critical. Now they're evolving into AI-compatible formats, and we're part of that.
Universities are another exciting case. Kennesaw State uses our data to help researchers brainstorm grant proposals. They analyze funding trends, political winds, and tech adoption before drafting proposals. The result? Smarter decisions, even on what not to pursue.
They analyze funding trends, political winds, and tech adoption before drafting proposals. The result? Smarter decisions, even on what not to pursue.
José Mauricio Duque: What I see most inspiring is how your system amplifies and enhances human intelligence.
Robert Caulk: Exactly. Analysts are more important than ever. The human brain is amazing at abstraction. AI is just another tool, like the pencil, printing press, or Google.
Used properly, AI elevates thinking. We've seen it firsthand. Our systems help executives and analysts make more informed decisions with more context.
José Mauricio Duque: Your platform enriches content with metadata: bias, geography, sentiment, and so on. How does this context engineering help enterprises make better decisions?
Robert Caulk: It's about building a higher-resolution picture. Details like source origin, bias, voice, and sentiment help LLMs interpret and treat the data appropriately.
A sensational tone? That should be treated differently than an objective one. All these signals help systems better understand and reason. We open source our models for extracting bias, entities, and more. They're used widely, and transparency is key for trust.
If a high-stakes system outsources trust to us, we owe them clarity. So our tools are open.
José Mauricio Duque: Let's talk about publisher protection. You've said scraping is legally risky and often loses nuance. How does AskNews's licensing-first approach solve that?
Robert Caulk: Scraping produces inconsistent data. No geo tags, missing context, snippets from unrelated articles. It confuses LLMs.
Legally, scraping is still being debated in court. But beyond legality, scraped data often lacks the nuance LLMs need. And it's bad for training. Our synthetic data is repetitive and robotic. It doesn't help an LLM learn language or creativity.
But it's great for inference, where you input data into a model that's already trained. That distinction is crucial. We're not training AI. We're helping it reason with curated facts.
José Mauricio Duque: Right. It's about better answers, not better writing.
Robert Caulk: Exactly. Training changes the model. Inference uses a trained model to generate output. We help with inference, not training.
José Mauricio Duque: With 75% of execs calling AI and predictive analysis top growth drivers, how do you see the relationship between publishers and AI platforms evolving?
Robert Caulk: I hope we help bridge that gap. We're already seeing platforms license content. Even competitors are doing good work in different niches.
Big publishers are landing huge deals. But what about the small ones? That's where Newstex and AskNews come in. Small publishers can't get boardroom deals, but their data is sometimes more valuable due to its diversity and specificity. Helping them get fairly compensated is a key part of our mission.
José Mauricio Duque: That's our goal too: to provide space for original, diverse voices. What role do partnerships with content licensing specialists like Newstex play in creating a new standard for how journalism flows into AI?
Robert Caulk: Licensing is like surgery. It has to be balanced, transparent, and built with trust. Scientists shouldn't be doing licensing. That's where Newstex comes in. You know the needs, the legal history, and how to build fair agreements. That's why we partnered.
About the author,
José Mauricio Duque
