I can hardly imagine scrolling through Google results as my only search tool anymore. It actually starts to frustrate me, and it wasn’t even that long ago that we all did it every day. It was the only option before we had LLMs to lean on for that.
A few months ago, I was sitting on the tarmac at JFK, already delayed, staring down a 75-minute layover at Charles de Gaulle. And if you’ve ever passed through there, that’s a worst-case scenario. I pulled up an LLM and asked: what terminal does my flight typically arrive at CDG? Then: what’s the transit like from that terminal to where my connecting flight leaves from? Then: give me a step-by-step guide to get through this airport as fast as possible.
It searched across social media, official airport sources, travel forums, and built me a routing plan with photos, in just a few minutes. Then I could move on to actually getting work done while waiting to take off. Or, if I was feeling particularly indulgent, booting up some reality TV or diving into a book. Either way, I had my time back, and I made my connection, to boot.
That’s what gets me excited about this shift in search capability. The actual experience of working the way you think at the pace you think it, not all of this theoretical, pie-in-the-sky talk about AI. Especially for an analyst: ask a question, iterate as needed, and move on. That’s the point for me, that feeling of not getting stuck for hours or days at the starting line.
I’ve had the really cool opportunity to be hands-on helping to build this kind of LLM functionality inside a real-world intelligence platform, letting users query an archive of reports in natural language. Getting to watch analysts adapt to these tools as we build them, all while remembering what it was like to be in their seat, has been one of the best parts of my job.
And once you have a tool where you can start exploring with whatever half-formed question is in your head, it changes how you work through your thinking. You don’t have to craft the perfect search term or logic structure using complex boolean language before you can begin. You just start typing and build from there. Goodbye “AND”. Goodbye “OR”.
As I sit in that proverbial seat, I’m finding I also think out loud more now and workshop ideas with LLMs whenever and wherever it suits me. Walking my dog. Driving home from work – or let’s be honest, stuck in traffic on The Beltway. I’ve always loved brainstorming with other analysts, talking through challenges, coming up with alternatives. Some of the favorite moments of my analyst days were sitting with a small group late at night, working through a tough problem on a whiteboard. Who doesn’t love an old-fashioned whiteboarding session? Now with an LLM, you have another avenue to do that. You put your thought process out there, typos and all, and iterate down to the question and answer you’re really trying to get at.
The speed at which analysts are adapting to this reality is what really amazes me. When I started intelligence work in the government, senior analysts talked about reading classified intel bulletins that arrived on reams of paper. They’d go through it by hand every day to synthesize. That wouldn’t even be possible today because the volume of data is exponentially higher. They of course didn’t have the expectation of getting anything faster than that. They worked at the pace of the technology they had.
But in the world we’re in today, there absolutely is that expectation that you’ll consume information at the pace it’s coming out – which in this case is at the speed of light. Or at least that you’ll try. This applies to people in leadership positions too. They’re consuming and expecting things that quickly. If you don’t figure out how to adapt in that reality, you risk getting left behind.
The Part That Makes People Nervous
And of course, the moment you start working this way, a specific anxiety shows up in intel circles. If those of us in intel are adapting, we all know what that means: our bosses will be too. Intelligence is at everyone’s fingertips. I think that’s scary to an intel and security team for a number of reasons.
It prompts several questions: Does this mean AI takes our jobs, when the Chief Security Officer (CSO) or the General has even more ‘real-time’ access to information? Will they cut budgets when leadership thinks they have an on-demand “AI analyst” only a click away? What makes an analyst valuable if the CSO can go to ChatGPT or have access to the same AI-based tools we’re using?
I share some of the same questions and concerns, and there are plenty of others we could highlight; the risk of making false assumptions or consuming misinformation, the fact that life safety is at stake. But I also believe that gatekeeping gave the appearance of keeping analyst jobs safe for a long time now. The fear was always that leadership, seeing raw information, would make their own assumptions, dismiss the analysts’ carefully constructed finished intelligence, and generate a million unnecessary questions.
But the age of the CSO and the General having this level of access is already here. We’ve been terrified of someone up the chain doing their own “intel work” for decades, and security and intelligence teams have always had to justify their value, especially in corporations where business functions are often under intense budget scrutiny.
What Hasn’t Changed
Information overload killed gatekeeping years ago; your CSO or General already has endless information at their fingertips. The problem was never access. It was too much access, too much noise.
The analyst’s value was never really gatekeeping that raw information from leadership. It was curation, synthesis, knowing what to ignore. And that’s still true, maybe even more so now.
Because your CSO can ask their favorite LLM anything they want. Ultimately, the value now is context and relevance, and it always has been. Humans know things that AI cannot. Your crisis manager knows why a certain plan won’t work because they drafted the after-action review the last time that approach was tried. The analyst knows the nuanced business thresholds for risk tolerance and how that applies to travel in a potentially dangerous location. The security manager has years of institutional knowledge and knows your CSO – what they like to see, how they like to see it, what’s the right thing to communicate at the right time in the organizational context that’s really going to resonate.
Those lived experiences can’t be found anywhere on the internet. Unfettered access to information can only go so far before the human needs to step in and say: no, THIS matters. THIS is what you need to know.
And that’s what relevancy actually means, not just having the latest information, but understanding the context and nuance of what your stakeholder needs to hear right now to make the decision in front of them. It’s knowing when the CEO cares more about reputational risk than operational details. It’s recognizing when that travel request is really about closing a deal worth millions, not just another trip to a potentially risky location. Staying relevant and in the loop means evolving with your stakeholder’s priorities, not just keeping up with the news cycle.
Where This Goes
I think there will be impacts to jobs and budgets as a result of this wave of AI tools. That’s coming for virtually every industry and most experts and large companies have said it explicitly. This is a lived experience already for many.
But I don’t think the analyst role goes away any time soon. We do, however, need to get clever about how we use these tools, how we show we can still do the work, that we can be credible and trustworthy while using them. That we have something to offer outside of what the tools provide. Our value proposition, if you will.
This isn’t fundamentally different from what we’ve always had to do though. Justify your budget, your people, prove your value and why you deserve a place in the room where decisions happen. Budget cuts, mission pivots, technology that promised to make us obsolete – we’ve seen versions of this before. The anxiety feels the same every time, that maybe this new tech development is finally the one that does us in.
The ones that come out the other side of this tech-driven shakeup figure out how to adapt, get business savvy, and make their case. They let their work speak for itself but more importantly they also speak for themselves about why their team deserves a seat at the table.
At Seerist, we’ve been putting our own LLM-based search through the paces with the kind of testing where we’re actively trying to find where it breaks down. If it can’t pass that kind of rigor, it doesn’t deserve a place in an analyst’s workflow and definitely shouldn’t be anywhere near a decision that could put lives or assets at risk.
Because let’s be honest: there’s a lot of that right now. Chatbots that make simple tasks harder, “intelligent” summaries that miss the point entirely, AI features no one asked for. Companies are racing to say they have AI without asking whether it actually serves anyone. It’s exhausting, and it makes the real advances harder to spot through all the slop.
And what our own testing has shown is pretty simple: the real shift isn’t the technology itself. It’s whether the technology serves the analyst’s judgment. When it does, you give the analyst time back to think critically, the space to explore the threads that matter, and the ability to focus on the analytical work that only humans can do.
I personally don’t want to go back to scrolling through endless search results on Google. I want to work the way I think with messy questions, half-formed hunches, and a lot of iterating until something clicks. That’s what we’re building: tools that sit inside that work and help analysts move at the pace they’re already expected to work. AI tools won’t replace the late-night whiteboard sessions, but they can get you there faster by stripping out the noise so you can get down to the real thinking.
