Enhancing Security Operations with Real-Time OSINT Intelligence

Part two of an edited transcript from an ISC West panel with Melissa Newberg, Global Head of Intelligence at Seerist, and Agnes Boros, Head of Product at Seerist. Part two focuses on the integration challenges between GSOC, strategic intelligence, and executive protection workflows, what an effective human-machine partnership looks like in practice, and why methodology has to come before technology in AI adoption. Part one was published separately.

How fast should an intelligence team make sense of a crisis?

AGNES: Now your operator has the alert. They know something real is happening. The next question is how fast you can make sense of it.

Your team doesn't have to read 100 articles to build the initial sitrep. The synthesis is already underway. And what surfaces alongside it, forward-looking scenario analysis, trajectory assessments, regime change implications, is what your leadership asks for in the next meeting. Having it already drafted, sourced, cited, changes the speed of your response entirely.

Now I want to talk about something honestly. Integration. Which is damn hard.

Why security team silos break down during a crisis

MELISSA: It's the part nobody wants to talk about because there's no clean answer.

Here's the version of this problem I see most often. You have a GSOC. You have a strategic intel team. Both are good at their jobs. And they are operating from completely different pictures of the world. Different feeds, different alerting thresholds, different ways of categorizing events.

Then something happens, Iran, or wherever, and the GSOC is calling up to the intel team asking "did you see this?" And the intel team is saying "yes, but we categorized it differently, and we sent something three hours ago." Nobody escalated because they didn't know they were looking at the same thing.

That's not a technology problem. That's a common operating understanding problem. When teams aren't building from the same foundation, same verified data, same sourcing standards, same definitions of what "significant" means, integration becomes a coordination tax that eats into the time you should be spending on analysis.

AGNES: Getting there is really hard, and the technology is usually not the hardest part. Most of you are not starting from a blank slate. Your GSOC has a workflow. Your strategic intel team has a workflow. Physical security has a workflow. Executive protection has a workflow. Each of those groups has spent years building their own tools, their own processes, their own tribal knowledge, their own definition of what a good answer looks like. They're not being obstructive. They've optimized for their own job. The problem is that a crisis doesn't respect those lanes.

We've had to learn this ourselves. This is hard for us too. We cannot work in silos on the product side any more than your teams can work in silos operationally. AskAnna is a good example. A tool like that can only be successful if it's built by a multidisciplinary team, intelligence, product, engineering, data science, all working from the same understanding of what the analyst actually needs. The moment those disciplines stop talking to each other, the product drifts away from the problem it was supposed to solve.

What we had to learn, and honestly, it took longer than I'd like to admit, is that you cannot solve the silo problem by forcing consolidation. People resist that, and they're right to. What you can do is make sure that even if the GSOC, the intel team, and the strategic function are doing different things, they are working from the same underlying picture. The same data. The same sourcing. So when leadership asks the question, everyone is giving an answer that comes from the same foundation.

How analysts and AI should actually work together

MELISSA: Agnes really teed this up. We have lived this integration through the development of AskAnna. It's built on that integration first and foremost: data science, engineering, intelligence, and product. You could have a really cool tool that doesn't at all serve the purpose of the end user, and the integration we've been talking about is what solves that.

Yes, we are building a solution at Seerist, but as Agnes said earlier, we are going into a time extremely quickly where everyone in this room is about to become a builder. So this isn't isolated to those building tools or SaaS solutions anymore.

AskAnna was quality tested and built with the end user in mind, really using AI for what AI is best at. It's not AI-written material. It's using AI to dig through and collate human-driven material, based on methodology, to provide a transparent answer the end user can trace and therefore have reasonable trust in. Built with credibility at the forefront.

This is where the analyst, the crisis manager, the GSOC lead, earns their keep. Machine scale handles collection, initial synthesis, continuous monitoring. Human judgment handles source credibility, organizational context, the "so what." Those aren't competing. They're sequential. The machine gets you to the starting line faster. The analyst takes it the rest of the way.

AskAnna is the interface for that partnership. An analyst types a question in plain language, what are the supply chain implications if the Strait of Hormuz shuts, and gets a narrative response. Mapped incidents, trends, supporting analysis, citations. Grounded in 12 months of curated human analysis. That's not replacing the analyst. That's giving them a head start that used to take hours.

The trust piece matters here too. Every response is sourced. The analyst can see exactly where the answer came from, evaluate it, push back on it, add context the system doesn't have. That's by design. The tool earns trust by being transparent about its own limitations. We love a no-answer at Seerist, because it means it's working as the walled garden it's designed to be. As we all go forward in this new world of agentic AI tools, these are absolutely crucial considerations we should be building toward.

Turning global risk into local decision advantage

AGNES: The last piece is where it gets specific to you. Making global risk local.

Everyone has access to the same open-source information now. What makes your intelligence function valuable is the proprietary context. Where your people are, what your risk tolerance is, what your decision cycle looks like. Your assets, your facilities, your supply chains, mapped against what's actually happening.

The platform isn't just telling you Iran is unstable. It's telling you what's happening in the places where you have exposure. Unique data becomes the differentiator. The system is only as good as the specificity you bring to it.

Why methodology has to come before technology in AI adoption

AGNES: Methodology before technology. You can move fast when you trust the foundation you're built on.

The organizations that are winning aren't the ones with the most data or the fastest alerts. They're the ones that have made deliberate decisions about what their foundation looks like. What verified means. What significant means. How they build from a common operating picture.

That's what we've tried to build at Seerist. And it's what we've seen work in the teams we partner with.

MELISSA: Many of us are dealing with matters of life safety in this industry, and you can't stake your reputation on just anything. If you're really going to try to achieve decision advantage, a huge part of that is trusting where you're getting your information to turn into intelligence. Methodology is the grounding of everything, especially as AI integrates into everything.

The bottleneck in intelligence has shifted. It's not access to information anymore, everyone has that. The bottleneck is interpretation, and the ability to act on it quickly, with confidence, at the right moment.

That's the job. That's what we're all here building toward.

Editorial note: The content above is drawn directly from a one-hour ISC West session led by Melissa Newberg and Agnes Boros. AI was used to condense and reproduce the audio into a readable article that captures the essence of the live discussion. The source recording is embedded at the top of this post.