Abstract Essence Ltd · Live — UTC
Abstract essence

An engine for finding niches the market hasn’t named yet.

Operations console · Live since April 2026 · Solo · UK
Hypotheses
58 tested
across all phases
Active
48 live
graduated · ranked · argument
Moves logged
1,349
since the engine began
Last admission
moments ago
—
Plate II — Topology of debate

Grading by disagreement, not consensus.

Every hypothesis the engine admits has been argued against by three independent voices — Anthropic, Google, xAI. The surviving claim is the one no model could break. Disagreement is the substrate; consensus is the residue.

That is what produces signal you can act on: not the model that is most confident, but the structure of objection between models that most disagree.

§ISurfaces — six ways into the system
  1. /01
    Hypothesis engine
    Every hypothesis the engine has tested, with corpus, council verdicts, and four-question Jones structure.Public · read-only mirror
    →
  2. /02
    Opportunity scanner
    Ten ingestion sources clustered into memos and gated against the Manifesto. Weekly cycle.Public · read-only
    →
  3. /03
    Daily digest
    Today's intelligence read — ten items scored on novelty, actionability, system relevance. Plus the proposal queue.Authenticated · Commander only
    →
  4. /04
    TraceSuture bundle
    Reference incident replay bundle — what a buyer receives at signature.Direct download · .tar.gz · 11 KB
    ↓
  5. /05
    Methodology
    How the engine grades. Adversarial debate, council verdicts, calibration curves.Public · living document
    →
  6. /06
    About the project
    Why a self-grading hypothesis engine exists. What it is not. Who is building it.Public · one page
    →
Plate III — Most recent admission

—

— — —
Plate IV — Manifesto
“A market sensor — before it is a product.”
Abstract Essence · founding thesis
Plate V — Identity Letterpress on bone paper · 2026