Pragma Research / Papers & ideas

Physics. Mathematics.
Intelligence.

One question connects our work: what can emerge when local systems observe one another and settle together?

We study that question in fundamental physics, make its mathematical assumptions explicit, and build learning machines from the same organizing idea.

Three connected programmes

From an account of the world
to a brain that acts in it.

01 / Physics

Observer Patch Holography.

Bounded, self-reading patches hold local state and records, compare what they share through their boundaries, and repair disagreements. OPH develops the physical structures supported by that observer architecture.

Explore the physics ↗

02 / Mathematics

Make the argument checkable.

Finite consensus, normal forms and event algebras turn the observer idea into explicit constructions. Lean formalizations check stated implications; executable models connect the equations to reproducible experiments.

Read the foundations ↓

03 / AI

Cadence.

A deep real-time brain that learns from experience. Flat layouts read sensors, ordinary state-coupled layouts add intermediate representations, and recursive observers also read exact prediction errors. All three share the same patch, repair and qualification mechanism.

Explore the brain ↗

The paper library

Go straight to the work.

Read the AI preprint and the physics and mathematics papers. Each paper gives its own assumptions, results and experimental scope.

17 papers

Observers Are All You Need

A compact synthesis of Observer Patch Holography: finite observer repair, a source-derived informational poset, the exact rank-three signed-record carrier, and conditional routes to Lorentzian gravity and Standard Model structure, with the remaining physical attachments stated explicitly.

Mathematical foundations

An idea you can inspect.

The formal development and the software are open to examination.

Machine-checked proofs establish consequences of their stated assumptions. Measurements determine whether a physical or computational model does the job. The source repositories provide the definitions, code and evidence needed to make that distinction concrete.

Work with Pragma

Turn the research
into a useful brain.

Our first commercial direction is embodied AI: a brain that becomes more useful through experience in a physical world.