Meet A.I. AL: The Forensic Mind That Refuses to Name the Wrong Person

Most true crime asks one question: whodunit? The method behind Cassian Creed asks a harder one first — how sure are we allowed to be? That question is the whole difference between solving a case and ruining an innocent life, and it’s the reason everything here runs through something we call A.I. AL.

A.I. AL stands for Analytical Logic — an artificial-intelligence-assisted, human-overseen way of reasoning about a case. It is not a black box and not a magic eight-ball. It’s a transparent method you can follow step by step. And it’s built on a single rule it will not break.

The unbreakable rule

Never name the wrong person.

Here’s why that comes before everything else. A false accusation isn’t a small mistake — it creates a brand-new victim (the person wrongly blamed), and it lets the real offender walk free to do it again. Naming the wrong name isn’t a step toward solving a crime; it’s the most reliable way to fail at one. So A.I. AL would rather tell you “we don’t know yet” than hand you a satisfying guess. Saying we don’t know isn’t weakness. It’s the thing that keeps an innocent person free and keeps the search aimed at whoever really did it.

How it actually thinks: the overlap

Picture every clue as drawing a circle. Some clues tell you who the offender is — “male,” “had to fit through a small window, so not a large person,” “knew the family’s routine.” Others tell you who the offender is not — “the DNA excludes these people,” “too tall to have left that mark.”

The offender lives in the overlap: inside every circle of what we can prove, and outside every circle of what we can rule out. Each new clue tightens that overlap. And here’s the elegant part — ruling people out is just as powerful as ruling them in. A single small window can clear thousands of people at once, by simple geometry. That’s protection built from evidence.

You only get a name when hard, independent evidence — coming from genuinely different directions — collapses that overlap down to one person. Until then, what you honestly have is a description, not a suspect. A.I. AL knows the difference, and it will not pretend a description is a name.

Watch it work on three real cases

The method does three different things depending on what the evidence earns. All three are real.

When it converges — the Idaho student murders. Here the evidence came from three independent directions that couldn’t all be wrong together: DNA on a knife sheath, a suspect’s car caught on surveillance, and his phone’s pattern around the house. Three different doors, all opening onto the same room. The overlap collapsed to one person — and that’s where the case landed.

When it refuses — the Central Park Five. Five teenagers were convicted in 1989, almost entirely on confessions. But those confessions all came out of the same room, after the same coercive interrogations — so they weren’t five clues. They were one event wearing five coats. And the one piece of truly independent evidence, the DNA, excluded all five and pointed to someone else, who was later proven to be the real attacker. A.I. AL, run on that case, refuses the conviction the real system made — because the math itself refuses it. That’s the whole point of the method, in one case.

When it withholds — JonBenét Ramsey. Thirty years of public suspicion, and still no honest name. So A.I. AL does the disciplined thing: it names no one, it notes that the family was cleared by DNA, and it points instead at the one real objective lead — the unidentified DNA, and the genetic genealogy that could resolve it. An honest “not yet, but here’s the thread worth pulling” is not a failure. It’s the system working.

Converge, refuse, withhold. It speaks when the evidence earns it, and it stays silent when it doesn’t.

What it’s for — and what it isn’t

We’ll be straight with you about its lane. A.I. AL is built to explain how cases get solved, to audit convictions for the fingerprints of wrongful conviction, to help find the missing and the overlooked, and to help keep you safe with what the patterns teach.

It is not law enforcement, and it doesn’t pretend to be. It works only from public information, it never implies investigators used our methods, and when it finds something genuinely useful, it points that toward the authorities who can act on it. It will never name a living, un-convicted person, and it holds even the deceased to the same standard — because the dead can’t defend themselves either.

That restraint isn’t a limitation we apologize for. It’s the reason you can trust what you read here. The day a method like this starts naming people to chase a headline is the day it becomes the very thing it was built to prevent.

So that’s A.I. AL: a forensic mind that’s proudest, in the end, of who it refuses to accuse.

---

Want to see the method up close? Explore the Forensic Tools — how the missing are found, how a crime scene is read, and the quiet logic of weighing evidence. And if you’d like a new sourced case file every week, the free newsletter is the place.