I spent years at the CIA, where the consequences of decisions were often measured in lives, not quarterly earnings.

In that world, the thing that mattered most was never the data. It was judgment.

Judgment is what tells you which data matter and which are noise. It’s what enables you to find threads between contradictory signals. It allows you to act when you have 60% of the picture, because waiting for 100% means acting too late.

Today, I’m watching the proliferation of AI across enterprises with concern.

General LLMs promise insights, but they often deliver semantic prediction dressed up as thinking. It’s pattern matching at scale, probabilistic guesswork packaged in a clean interface.

And yet, business leaders are being asked to make critical decisions based on it. General LLMs deliver confidence without rigor, and recent research shows that even smart people tend to trust AI that sounds sure of itself.

As I'm sure my former colleagues would firmly agree, intelligence is not just information. It is information collected with purpose, analyzed with rigor, challenged with structured dissent, and delivered with insight and context. There is an entire discipline (decades of methodology) behind turning raw data into something a decision-maker can trust.

AI has skipped all of this. It went straight from “we have a lot of data” to “here’s your answer” and left out the reasoning layer.

My co-founder, Ari Popper, and I set out to build AI systems capable of producing decision-grade intelligence, using a discipline we call "Judgment Engineering."

We didn’t start with the technology. Like any good CIA analyst, we started off with the single most important part of the process. We sought to ask the right question:

"What does it look like when experts reason through a hard problem?"

In other words, how do real, human experts think?

We studied intelligence analysts, scenario planners, futurists, social and behavioral scientists, and other experts. We delved into how the best strategic thinkers in the world structure ambiguity into something actionable—their reasoning loops, heuristics, frameworks, ontologies, and more.

And then we engineered it.

Judgment Engineering isn’t about better prompts; it’s about architecting the cognitive process itself. It is the practice of building AI systems that encode how experts reason to arrive at judgments (structured techniques, traceable logic, assumptions checks, etc.).

At Dreamport.ai, we believe the next era of AI isn’t about making machines faster.

It's about engineering AI that decision-makers can actually rely on.