Point it at anything — a market, a decision, a policy, a document, an institution. The engine reads the situation sixteen different ways at once, checks its own conclusions for bias, and only then commits to an answer: what's really going on, and the first thing to do about it.
Tier 1 — the material ground. Six parallel readings of what is verifiably true: who controls what, who pays and who benefits, who's actually connected to whom, how this situation came to be, and what effects are already documented. No interpretation yet — just the ground.
Tier 2 — the structural reading. Nine more readings on top of that ground, looking for what the surface story hides: what the dominant player can't admit, what the challenger story gets wrong, and — the load-bearing question — what both sides avoid saying, because that's where the real opening usually lives.
Tier 3 — validation, then the verdict. Before any conclusion is allowed, an adversarial pass attacks the first fifteen readings: where did the analysis see what it wanted to see? Only after that correction does the engine commit to its final synthesis — the moment of breakthrough it sees, and the single first thing to build or do.
The engine presents what the structure demands. It never tells you what to want — that part stays yours.
Each run is a real, full-depth analysis — the free tier limits how much you feed it and how often, not how carefully it thinks. Longer documents, deeper multi-run studies, and higher volume come with paid tiers (beta users keep beta pricing indefinitely).
Free with sign-up — every account includes the engine, rate-limited on the free tier as listed above.