Introducing Radiant Intel

2/4/2025

Today we launch Radiant Intel, an agentic artificial intelligence platform that takes complex global data and turns it into clear, actionable insights, helping our clients confidently navigate a difficult world. Our platform features expert-curated proprietary data, AI agents that shape analysis based on human expertise, and a fully customizable reporting solution that integrates seamlessly into any enterprise. We consistently hear from clients that they don’t want more data. They want answers that directly translate to decision advantage. That’s where Radiant Intel comes into play to deliver actionable and customizable intelligence that far exceeds the capabilities of off-the-shelf foundation models thanks to difficult-to-acquire data and expertise. The upside to our clients lies in deploying domain expertise to go far beyond the underlying capabilities of the model.

AI’s enormous upside in enabling domain expertise at scale

The role of AI agents doesn’t replace analysts. It enables them to focus on higher value work for the enterprise, making domain expertise scalable. Analysts need to learn on the job. They don’t have infinite time. So, what do they do? They heavily weight reporting with what’s most proximate, highest value, and most likely to get accepted within the publishing schedule. Senior analysts may review work and suggest edits, but they don’t ever say “You missed these 300 important sources, read them all by COB tomorrow”.

In an ideal world, an analyst has the time to do that reading. But our clients don’t have infinite time. And that is the primary limitation in the dominant analysis paradigm, in both the government and the private sector: good analysis doesn’t scale. Analysts can’t go chasing sources that they aren’t confident will be worth their time. We have the domain expertise and technical capability to handle that curation for you.

Radiant Intel advantage #1: expert-curated, proprietary database

Many of the most interesting open-source artifacts are quite complicated to get at. We know from our careers what constitutes high-signal sources. News and social media are generally downstream of actual intelligence (with the occasional leak being the exception). Compared with our expert curation, the cost of storage and processing associated with big data is simply not competitive. Internet search lacks the kind of comprehensiveness that our clients require. Some sources are interesting but far too long to digest properly for robust manual analysis (lawyers may read every page of a 150-page law, but analysts do not). Others are boring in isolation, but every so often they collectively yield great insight into a government’s motivations or intentions. In a world of data overload, we solely rely on information that will be useful.

The remedy to AI’s “garbage in, garbage out” mantra is “quality in, quality out.” Our database features primary sources that are authoritative indicators of what governments say and do. We process everything that governments publish to triage only the most important information and separate signal from noise.

Radiant Intel advantage #2: analyst expertise encoded in AI agents

The Radiant engine takes unscalable human analysis and makes it scalable. Our AI agents are coded with the expertise that a human with infinite time would use to evaluate the entire universe of top-quality primary sources. Our team spends their time thinking about the rules that make the world turn, and the system draws on our proprietary database to generate analysis based on those human-coded rules.

In our case, rather than asking the model to produce an open-ended analysis of a complex issue, we set up significant “intellectual scaffolding” to shape the model’s output based on human expertise. The trade-off is that this approach requires decades of expertise in a particular domain. Currently, our platform consists of a few hundred AI agents. I expect we will have thousands in the next few months, and perhaps tens of thousands in the coming years as we expand our operations into new areas and enable our clients to add their own to their client.radiantintel.com enterprise domain using our engine.

The automated part doesn’t replace the human analyst. It augments the work and redirects human focus towards the really tough intellectual questions that require decades of domain expertise to answer. What are the telltale signs of an invasion of Taiwan? How did the pandemic alter the world’s ability to detect and respond to novel pathogens? What does the surge of populism mean for global politics? As a consumer of analysis, I want experts to spend their time thinking about those questions and meeting with people who have unique insights.

Radiant Intel advantage #3: customizable enterprise integration

Another thing we’ve heard from clients is the frustration with many different sources of content that don’t integrate in their existing workflows. That’s why Radiant Intel isn’t just a collection of intelligence briefings—it’s a platform and place of work for any team charged with understanding the world for their organization.

Our clients benefit from our insights to help set their strategic agenda. But they can also direct the Radiant Intel engine at challenges and opportunities that are already on their agenda. Our enterprise solution features customizable output that can meet any team’s reporting needs. Enterprise clients get the benefit of our proprietary data and AI agents while also being able to incorporate their own data and expertise into their analysis on a dedicated radiantintel.com client domain. Data privacy and cybersecurity are core values of Radiant Intel, but they also enable clients to get the best possible analysis by combining our technology with their privileged insights that don’t exist outside of the enterprise.

The implications of these three advantages are profound for our clients.

• They move from an industry paradigm that is unscalable to one that is scalable to the entire universe of high-quality sources.

• Because they benefit from the whole universe of high-quality sources, they derive novel insights that the old paradigm lacks.

• They have the confidence that they are collecting and analyzing everything that matters.

• They learn more about what actually constitutes signal for their organization.

• They spend all their time focusing on high-value strategic initiatives, and no time on the more mundane aspects of their job.

This is an incredible moment in the AI era. Jensen Huang, the CEO of Nvidia, recently said “if the marginal cost of intelligence was practically zero, there are a lot of things that you would do now that you wouldn’t have done otherwise.” Our request to our clients is to let us deliver intelligence so you don’t have to make choices about what you collect or how you analyze it, all while spending more time on the KPIs that matter most to your organization.

Michael Frank