Unlocking the Power of Generative AI in Dow Jones with Ingrid Verschuren
Fresh out of the studio, Ingrid Verschuren, Executive Vice President of Data and AI and General Manager, EMEA, at Dow Jones, shares her journey from manually indexing news articles at Reuters to leading Factiva’s transition into the era of generative AI. She explains the mental models and frameworks behind how Dow Jones uses AI to empower businesses with reliable data and insights and discusses the recent partnership with Google’s Gemini model and explains how Dow Jones works with the other content publications, navigating the complexities of trust and business model innovation. Last but not least, Ingrid explains what great would look like for Factiva in the age of AI.
"So, we are very conscious of the fact that we license the content from other publications. And as I mentioned previously, we do that through licensing agreements. We are transparent with the publishers about what happens with their content. We ensure that they are being fairly compensated for the content that we use. But, as a result, when we started talking about how we wanted to use Gen AI, we felt that we had an obligation to go back to publishers and ask for additional licensing rights. Part of that, I think, is driven by the fact that Dow Jones is a publisher. We are a publisher ourselves. We are very protective of our content. We want to make sure that we understand what's happening with our content: Where is it going? Who is using it? And we want to be fairly compensated for it. So, if that's one of our core principles, then we also want to make sure that we treat all the other publishers the same. One of the challenges has been that we had to go back to all publishers and ask for additional Gen AI licensing rights. The positive of that is that the content now available through Gen AI is fully licensed content. All publishers have given us permission to use their content for this specific use case." - Ingrid Verschuren
Profile
Ingrid Verschuren, Executive Vice President, Data and AI and General Manager, EMEA at Dow Jones. (LinkedIn)
Here is the edited transcript of our conversation:
Bernard Leong: Welcome to Analyze Asia, the premier podcast dedicated to dissecting the pulse of business technology and media in Asia. I am Bernard Leong, and generative AI has transformed how information flows globally. How does a business intelligence platform such as Factiva navigate in the age of AI? With me today, is Ingrid Verschuren, Executive Vice President of Data and AI, and General Manager, EMEA, Dow Jones. Ingrid, welcome to the show.
Ingrid Verschuren: Thank you very much for having me. I'm super excited to talk about this today.
Bernard Leong: Yes, I'm very excited to have this conversation. Previously, I spoke to Joel Lange on the show, and we were talking about AI and compliance. This is much more exciting because we're talking about Factiva, which is a platform that a lot of businesses use to flow financial information. To begin, we always like to talk about the origin story of our guests. So, how did you start your career, which eventually led you to your current role at Dow Jones?
Ingrid Verschuren: Absolutely, I’d be happy to share. My career path is an interesting one. I hold a master’s degree in Latin American Studies with a minor in Business Management. Originally from the Netherlands—something you can probably tell from my accent—I moved to Spain years ago. At the time, finding a job that aligned with my degree was challenging.
I eventually secured a position at Reuters, which later became part of Dow Jones. My role involved manually indexing news articles in German, Dutch, Spanish, and Portuguese. While it may seem routine, this job offered incredible variety. One moment, I’d be researching the structure of Burmese names; the next, I’d delve into technology solutions or budgets. It was a dynamic environment that allowed me to explore diverse topics—something that has kept me engaged throughout my career.
Bernard Leong: Given you have such a long tenure with Dow Jones, what valuable lessons can you share from your career journey?
Ingrid Verschuren: I’d say there are three key lessons I’ve learned over the years. First—and most importantly—is the value of working with people you like. We all spend a significant portion of our lives at work, and being surrounded by smart, enjoyable colleagues makes the experience far more fulfilling.
The second is finding purpose in your work. The purpose is incredibly motivating. At Dow Jones, our mission is to deliver trusted journalism, data, and analysis that empower people to make informed decisions. We hold the world accountable and inform people with facts—a truly meaningful way to spend your career.
Lastly, never hesitate to say yes. Opportunities often come disguised as challenges, and it’s easy to doubt your ability to succeed. Early in my career, I made a habit of embracing these challenges, and it has paid off immensely.
Bernard Leong: Today’s discussion focuses on Factiva in the era of generative AI. For those unfamiliar with the platform, could you introduce Dow Jones and its Factiva business intelligence platform? What role does it play in supporting businesses with reliable data, insights, and risk management solutions? When I was the Head of AI and Machine Learning for AWS, particularly when working with financial services clients, Factiva was consistently a key part of the conversation.
Ingrid Verschuren: For those who don’t know, Factiva is a business intelligence platform best described as a comprehensive news aggregation database. It features over 33,000 sources in 32 languages, making it incredibly diverse. One of the core strengths of Factiva is its commitment to licensing these sources directly. We work closely with publishers, securing their permission to include their content in our database—and we compensate them for it. This ensures that the information users access is trustworthy and reliable.
Another standout feature of Factiva is its access to content that isn’t freely available on the web. Many news articles are locked behind paywalls or subscription models, but Factiva provides a gateway to that critical information. Additionally, its extensive archive contains billions of news articles, enabling users to conduct historical research with ease.
When you combine this vast repository with corporate data—such as details about companies and executives—it empowers professionals across industries. Whether you’re in government, academia, legal practice, or consulting, Factiva provides the tools for deep, informed research.
Bernard Leong: Before this interview, I delved into your background and discovered that you’ve had a front-row seat to the evolution of Factiva. You began as a news indexer at Reuters, manually tagged data, progressed through the digitization era, and now lead in the age of AI. It’s quite rare to meet someone who has witnessed and shaped such an extensive transformation.
What core values or principles have remained constant for Factiva throughout its journey? And how have these values shaped the platform’s adaptation to new technologies like generative AI?
Ingrid Verschuren: That’s an interesting observation, and I often reflect on it. To provide some context for your audience, my early role involved manually reading news articles in various languages and tagging them with metadata. For example, we’d identify an article as covering a merger, note the companies involved, specify the industry and detail the countries mentioned. This process was entirely manual at the time.
As the volume of news articles grew, it became clear that manual tagging wasn’t scalable. Adding more people wasn’t a sustainable solution. So, we started leveraging early forms of natural language processing (NLP) to automate the process. Within four to five years of my starting the job, much of the manual work had been fully automated. Yet here I am—proof that automation doesn’t replace people but rather shifts their focus to higher-value tasks. I think this experience has helped dispel fears about AI and automation for many.
Today, Factiva processes between 600,000 and 700,000 news articles daily, which necessitates constant innovation. We’ve consistently sought smarter ways to manage and utilize this vast data. For instance, last year, we launched Factiva Semantic Search, which made searching within the platform much more intuitive and user-friendly, even for those without deep expertise in information retrieval.
Now, we’re taking the next step with generative AI. One of our latest innovations is GenAI summarization. It helps users quickly make sense of Factiva’s vast information pool by summarizing search results. You can pose a question, and the system delivers a concise summary of the relevant results. This feature not only saves time but also provides clarity about the sources behind the summary. For example, the system identifies which articles the summary is based on and allows users to read the full articles if needed. This transparency ensures that the information is traceable, citable, and auditable.
These principles—transparency, trust in content, and maintaining the integrity of our sources—have been with us for over 25 years. They remain the foundation of Factiva, even as we embrace cutting-edge technologies like generative AI.
Bernard Leong: That’s a fascinating perspective. Can you explain how AI, beyond just automation, is transforming the way businesses approach risk management today?
Ingrid Verschuren: Sure. At its core, AI and generative AI certainly help with automation, enabling businesses to operate more efficiently. But for us, the real power of these technologies lies in their ability to address complex customer problems.
Our approach to introducing new technology has always been customer-driven. We don’t adopt new tools for the sake of innovation alone. Instead, we start by listening to our customers, understanding their challenges, and then developing solutions that genuinely add value.
Beyond automation, AI and generative AI are revolutionizing how we process and interpret vast amounts of unstructured information. Historically, this has been a major challenge. These technologies now enable us to make sense of such data at an unprecedented speed, helping businesses uncover insights that were previously difficult to access.
Bernard Leong: Adopting generative AI must come with its challenges. Were there any significant hurdles or opportunities you encountered as you integrated this technology into Factiva?
Ingrid Verschuren: One of the key challenges is rooted in the fact that we don’t own much of the content we work with—we license it from other publishers. As I mentioned earlier, we have licensing agreements that ensure transparency with publishers, fair compensation, and clarity about how their content is used. When we began exploring generative AI, we realized we had an ethical obligation to revisit these agreements and secure additional licensing rights specific to this technology.
This was especially important because Dow Jones is also a publisher, and we are deeply protective of our own content. We want to ensure it’s used responsibly, that we understand where it goes and how it’s utilized, and that we’re fairly compensated. Extending the same principles to other publishers was a natural decision.
While this presented a challenge—requiring us to renegotiate rights with all our publishing partners—it has resulted in a significant positive outcome. The content now accessible through generative AI is fully licensed, with explicit permissions granted by all publishers. This ensures we uphold our values of transparency, fairness, and trust, which are fundamental to our operations.
Bernard Leong: For example, how does Dow Jones use generative AI to safeguard against misinformation or detect anomalies? Could you share some examples of these technologies in action?
Ingrid Verschuren: The key lies in the quality of the input data. It starts with ensuring that we use trusted, reliable information. This is why our licensing process is so critical. We carefully vet the publishers we work with, verifying that their content is dependable. Our global licensing team, familiar with media landscapes across regions, applies human judgment to determine whether a publisher is trustworthy. If so, we establish a licensing agreement.
This rigorous process ensures a reliable data foundation, which helps mitigate risks like misinformation. That said, it doesn’t entirely eliminate issues like hallucination in generative AI models—though it significantly reduces them. Another critical component is constant testing. Generative AI systems, particularly those relying on prompts, can be highly sensitive. Even a small prompt adjustment can disrupt the output. That’s why we engage in extensive and ongoing testing. Launching a solution isn’t the end of the process—it’s the start of continuous optimization.
Bernard Leong: What’s one thing you know about data and AI in Factiva that very few people do?
Ingrid Verschuren: I know so much about it, it’s hard to pick just one!
Bernard Leong: I would love to hear all of them. [Laughs]
Ingrid Verschuren: Let me share two things. First, while Factiva is widely known for its news aggregation, it also offers a wealth of corporate and executive information. We cover over 40 million companies and more than 80 million executives. Combining this dataset with our extensive news repository creates a powerful resource for our clients, whether they’re conducting market research or due diligence.
Second, from the very beginning, we’ve tackled multilingual challenges. Even 25 years ago, when we managed fewer than today’s 32 languages, we had more than just English to contend with. Back then, there was no Google Translate or similar tools. We developed a system to tag articles with metadata, enabling users to search for specific topics, like mergers, regardless of the article’s language. This approach allowed someone who didn’t speak Japanese or Chinese, for example, to find relevant articles about mergers in those languages.
Over time, we built on this system. When machine learning, particularly supervised learning, emerged, we were ahead of the curve because we had years of annotated data. This foundation has allowed us to continue innovating, making Factiva a leader in multilingual and multilayered content management.
Bernard Leong: One of the significant challenges for many business owners, especially within large enterprises, is deciding whether to build their own large language models (LLMs) or adopt a retrieval-augmented generation (RAG) approach using existing enterprise LLMs. What was the thought process behind Dow Jones partnering with Google and adopting the Gemini models for use in Factiva?
Ingrid Verschuren: That’s a great question, and I think it helps to break it into two parts: first, choosing the platform, and second, determining how to use it. Starting with the latter, one of the key advantages of using a RAG model is the high level of control it offers over both the input and output.
With this approach, we can use all the licensed Factiva content for generative AI, store it in a vector database, and allow the RAG model to draw from that controlled dataset. This control is crucial for several reasons. First, it ensures that the answers provided are derived from reliable, licensed content. Second, it enhances transparency with publishers. We can trace each piece of information back to its original source, identifying which publication contributed to the output. This is critical for maintaining trust with our publishing partners.
Moreover, this setup allows for flexibility. If a publisher decides to end their partnership with us, we can remove their content from the system entirely. By contrast, if we had opted to merge everything directly into a standalone large language model, the content would become embedded and effectively irreversible. This level of control and transparency was a significant factor in choosing the RAG approach.
As for partnering with Google, it was a natural choice. We already had a strong working relationship with them and were utilizing various Google Cloud solutions. Their Gemini model stood out due to its multilingual capabilities, which are essential for Factiva’s support of 32 languages. The collaboration has been incredibly positive, with Google providing strong support throughout the journey. Their platform’s capabilities aligned perfectly with our needs, enabling us to continue delivering high-quality, reliable insights to our clients.
Bernard Leong: Based on the partnership, how will Google help Dow Jones enhance its AI-powered solutions within the Factiva platform?
Ingrid Verschuren: The partnership has been incredibly collaborative, and it has allowed us to address some key challenges effectively. For example, one critical aspect we evaluated was latency—ensuring the system could process the vast amount of content we handle efficiently. Google’s tools met our requirements in this regard, which was an important factor.
Another crucial element was their multilingual capabilities. With Factiva supporting 32 languages, having a robust multilingual model was non-negotiable. Additionally, Google’s platform offered the ease of use we needed, making the implementation seamless and efficient.
Bernard Leong: You mentioned traceability earlier—being able to track the provenance of data. This is particularly relevant when licensing expires and certain content needs to be removed. What specific benefits do businesses gain from the transparency features embedded in Gemini-powered tools?
Ingrid Verschuren: The biggest benefit is auditability. Many of our clients, whether they are in consultancy or compliance departments, need to document their research thoroughly. They must show where the information originated and justify the decisions they’ve made.
With Factiva powered by generative AI, we ensure that all content is licensed, removing any potential copyright issues. Furthermore, we make it clear where each piece of information comes from, which is critical for maintaining trust and integrity in decision-making processes.
Bernard Leong: I often tell CEOs that hallucination in generative AI is more of a feature than a bug. But it’s also a concern for many. Given Factiva’s reputation as a trusted resource, how do you address hallucination issues in your generative AI products?
Ingrid Verschuren: While I’m not a technologist, I can speak generally about our approach. The key lies in continuous testing. Generative AI models, particularly those reliant on prompts, can produce hallucinations if the prompts aren’t fine-tuned. When we identify such issues, we make prompt adjustments to prevent them. However, it’s an ongoing process—solving one issue can reveal another. That’s why rigorous and consistent testing is essential. It’s not a one-time task; it’s a continuous effort.
Bernard Leong: How would you advise your customers to use generative AI tools to remain agile, particularly in response to new compliance requirements or regulatory changes?
Ingrid Verschuren: Flexibility is essential. Rather than focusing on a single-purpose solution, we recommend building holistic systems or modular platforms that can adapt to multiple needs. By doing so, businesses can quickly adjust to new compliance requirements or regulatory changes without overhauling their systems.
From a data and content perspective, structure is equally critical. The more structured your data and content are, the easier it becomes to manage and adjust as needed. Even though Factiva processes unstructured text, we’ve maintained a highly structured approach to how it’s ingested and fielded. This allows us to quickly identify and exclude any information that’s no longer permitted for use. Ultimately, it’s all about creating a flexible, structured system that can evolve with changing needs.
Bernard Leong: Generative AI is advancing at a rapid pace—nearly every week, I’m chasing the latest innovations. For Dow Jones, which continues to push boundaries in business intelligence, what do you see as the next big advancement in AI for risk management and decision-making? Not to predict the future, but what trends or developments are you watching closely?
Ingrid Verschuren: If I could predict the future, that would certainly make things easier! But in the near term, one area I’m particularly excited about is the potential to merge structured and unstructured data effectively. Historically, these have been treated as separate entities—structured data being clean and neatly organized, and unstructured data being far more chaotic, like raw text.
The next major step is to combine these two and apply advanced search capabilities across the integrated dataset. By doing this, we could create an incredibly intelligent and versatile platform that empowers users to extract insights more holistically and efficiently.
Bernard Leong: What’s one question about AI and data in Factiva that you wish more people would ask?
Ingrid Verschuren: A lot of conversations tend to focus heavily on the technology itself, but I think a more intriguing question is: What is the importance of the human element in this process? It’s something we’ve reflected on at Dow Jones, and a few years ago, we coined the term "authentic intelligence." The idea is that artificial intelligence alone isn’t enough—you still need human intelligence and judgment.
For instance, humans are essential in determining what constitutes a reliable input source or in evaluating the output of a model. Machines, as advanced as they are, can’t yet decide what "good" looks like when it comes to nuanced or complex questions. Our approach is to let machines handle tasks they excel at, such as processing vast amounts of data quickly while freeing up humans to focus on deeper, more investigative work that adds real value for our clients.
Bernard Leong: So, it ultimately comes down to human judgment. That leads me to my traditional closing question. What does "great" look like for Factiva in this new age of generative AI?
Ingrid Verschuren: "Great" for Factiva means staying true to our core principles. It’s about combining structured and unstructured data in a seamless way to unlock powerful insights while continuing to rely on trustworthy information. Transparency with our publishing partners is also crucial—ensuring we compensate them fairly and maintain their trust. If we can achieve all of this while pushing the boundaries of what’s possible with AI, then we’ve done more than just good—we’ve done great.
Bernard Leong: Ingrid, thank you so much for joining the show and spending quality time discussing Factiva and its integration with generative AI. Congratulations on the partnership with Google on the Gemini model—I’m looking forward to seeing some of its features in action. Before we wrap up, I have two quick questions. First, are there any recommendations that have inspired you recently?
Ingrid Verschuren: I’ll go with a book I read a couple of years ago, which remains the best I’ve read since then: A Little Life [by Hanya Yanagihara]. It’s a deeply emotional and, admittedly, very depressing book, but it’s beautifully written. The anxiety and struggles of the characters practically seep through the pages. Despite its sombre tone, it reminded me of the importance of cherishing life and enjoying its moments—an interesting takeaway, given the story itself leans in the opposite direction. That would be my recommendation.
Bernard Leong: That sounds like a fascinating read. Since we’ve just come off the U.S. presidential elections, there’s certainly no shortage of relevant themes—but let’s save that for another time! My final question: how can my audience find you?
Ingrid Verschuren: The best way to find me is on LinkedIn. I’ve also done a couple of interviews, including one from a few years ago on YouTube that still seems to resonate with viewers. Currently, I’m focused on engaging with the publishing community, as collaboration is key. We believe that by working together as a community, we stand a much better chance of being fairly compensated for our content and ensuring its proper use.
Bernard Leong: Thank you so much, Ingrid. To our listeners, you can find us on all major podcast platforms, including YouTube, and Spotify. Don’t forget to subscribe to our newsletter, either on LinkedIn or through our website. Ingrid, it’s been a pleasure having you on the show.
Ingrid Verschuren: Thank you—it’s been great to be here.
Bernard Leong: Looking forward to continuing this conversation in the future.
Podcast Information: Bernard Leong (@bernardleong, Linkedin) hosts and produces the show. Proper credits for the intro and end music: "Energetic Sports Drive" and the episode is mixed & edited in both video and audio format by G. Thomas Craig (@gthomascraig, LinkedIn). Here are the links to watch or listen to our podcast.
