The hypothesis layer
for interdisciplinary science.

Cross-disciplinary breakthroughs are delayed by decades not because the insight is missing — but because the recognition precedes the vocabulary. The Consilience is built for that moment.

Not similarity

Every retrieval system matches words. The Consilience matches underlying structure — the recognition beneath the vocabulary.

Not generic

Generated entirely from your own sources. Cited, traceable, in your voice. Not from the internet.

Not a chatbot

A private knowledge graph that deepens with everything you ingest and compounds across your entire intellectual history.

Watch the video Launch App →

Web  ·  Mobile

Think Deeper. Write Truer.

Who It’s For

The Interdisciplinary Researcher
Working across physics and adjacent disciplines. Whose most important recognitions arrive before the vocabulary to express them exists.
The Deep Thinker
A philosopher, founder, or writer whose thinking crosses disciplines and whose output needs to be grounded in what they actually know — not what the internet says.
The Research Institution
Groups building shared knowledge infrastructure at the frontier. Multiple researcher lenses. One domain corpus. Collective vocabulary that compounds from real usage.
Early access
Currently open to selected researchers in physics and interdisciplinary science. Request access →

The Problem

01
The vocabulary wall
A neuroscience paper on “default mode network deactivation” and a contemplative text on “dissolution of the observer” point at the same structure. No retrieval system sees it. The words share nothing. The recognition is identical.
02
AI that generates from the internet, not from you
Current AI generates from the world’s data. Your most important work requires generating from your own — from what you have actually read, thought, and built over years.
03
Thinking that accumulates, never compounds
Years of notes, highlights, voice captures, and realisations scattered across a dozen apps. The connection that matters at the moment of writing exists somewhere in your own history. No system holds it together.

Where We Sit

The moment before the discovery
The recognition that precedes the proof.
Every discovery has two moments: the recognition, and the proof. Today’s tools — formal verifiers, automated labs, literature synthesis — handle the proof extraordinarily well. The recognition has no tool. The Consilience is building it.
Shannon · Boltzmann · Lotka · Volterra
The delay was never the experiment.
Boltzmann’s entropy described how the brain minimises surprise a century before neuroscience formalised it. Katalin Karikó’s mRNA insight was dismissed for decades before it became the basis of a vaccine. Fleming noticed the penicillin mould fifteen years before anyone acted on it. The insight always existed. The infrastructure to surface it did not.
Schrödinger · Bohm · Heisenberg · Oppenheimer
“Quantum theory will not look ridiculous to people who have read Vedanta.”
Not poetry. A response to formal results — the measurement problem, non-locality, the observer-observed boundary — that classical vocabulary could not hold. Schrödinger read the Upanishads while developing wave mechanics. Bohm developed the implicate order through dialogue with contemplative philosophy. The structural resonance they sensed is precisely what The Consilience is built to surface.
Now · Next · Horizon
Physics and contemplative science first. Neuroscience next.
Current focus: quantum physics and its structural parallels with contemplative science — where the historical evidence is strongest and the tooling most absent. Next: neuroscience, where correspondences can be checked against empirical data. On the horizon: a domain-specific model trained from the knowledge graph, and ongoing research into foundational architectures beyond similarity-based retrieval.

Generation Modes

Synthesis
The original insight at the intersection of your traditions. Surfaces what neither field has stated — grounded entirely in what you have ingested.
Brainstorm
Wide-net lateral thinking across your knowledge graph. Surfaces surprising connections you would not find yourself.
Gap Finder
What your sources cover that you have not yet articulated. Your open questions, ranked by how much evidence already exists in your graph.
Parallel Lens
Your own thinking alongside your sources. Where you align, where you diverge, where you have gone further than anything you have read.
Tradition Verdict
How each tradition in your knowledge graph answers the same question. Structured contrast, grounded in your ingested sources only.
Essay & Book Chapter
Long-form writing in your voice. Structured, argued, fully cited — built from your knowledge graph, not the internet.
Journal
A private thinking space that responds from your own corpus. Reflection grounded in what you actually know and have experienced.

The Architecture

Built on the distinction between surface similarity and structural recognition. Three layers of representation. One knowledge graph that learns from use.

Layer 01
Three-level embedding
Every source is represented at three levels: surface vocabulary, conceptual structure, and a vocabulary-neutral essence abstraction. The essence layer is where “quantum vacuum” and “Brahman” become retrievable together.
surface · conceptual · essence
Layer 02
Hybrid retrieval
BM25 sparse search, dense vector search, knowledge graph traversal, and a reranking pass — run in parallel and fused. The retrieval is not a keyword search. It is a structural query against everything you have ingested.
BM25 · dense · graph · rerank
Layer 03
Living knowledge graph
Auto-populated during ingest. Recognition clusters form as sources connect. Vocabulary bridges link terms across traditions. The graph compounds with every source added — it does not reset or summarise.
auto-populated · multi-hop · compounding
Layer 04
Confirmation as training signal
Every time a researcher confirms or refines a cross-tradition connection, that judgment enters a training dataset. Over time, these confirmations calibrate a domain-specific model that learns this field’s vocabulary from expert behaviour — not from a generic prompt.
confirmation · vocabulary map · dataset
Layer 05
Passive intelligence distillation
Four training datasets accumulate from real usage: query-relevance pairs, essence translation pairs, entity-relation examples, and vocabulary equivalences. When thresholds are reached, a domain-specific adapter is trained — a model that improves retrieval precision in a specific field.
Mistral 7B · QLoRA · four datasets
Layer 06
Grounded generation only
Generation is grounded entirely in what has been ingested. When the knowledge graph cannot support a claim, the system surfaces the gap rather than filling it with world knowledge. Every output is cited and traceable to a source.
grounded-or-silent · cited · traceable
Research directions  ·  Active
Structural correspondence detection
Moving beyond similarity-based retrieval toward structure-mapping: detecting when two bodies of knowledge have the same relational form beneath different vocabularies. Informed by structure-mapping theory (Gentner, Forbus) and causal representation learning (Schölkopf).
Domain-specific small language model
A domain adapter (Mistral 7B + QLoRA) trained from the knowledge graph’s accumulated confirmation data. Not a general model — a specialist that understands one field’s cross-tradition vocabulary because researchers in that field trained it.
Foundational architecture research
Actively researching the architectural frontier: verifier-centric systems, world models (LeCun, JEPA), causal representation learning, and neurosymbolic approaches. The long-horizon goal is a representation layer where structural correspondence is decidable, not approximated.

Research & Writing

On the nature of intelligence, the limits of current AI architecture, and why the most important gap in scientific discovery has no tool yet.

Essay  ·  2026
Intelligence Was Here Before We Were
On cognition, consciousness, and what the architecture of mind tells us about the architecture of machines. A mycorrhizal network, a slime mould, Ramanujan's dreams, and the question of whether we have been measuring intelligence by the wrong instrument entirely.
Read essay →
Essay  ·  2026
After the Transformer
AI solved an 80-year-old maths problem. It found 31 new planets. Here is what that actually means — and where the real gap in scientific discovery lives, the one that no amount of scaling, robotics, or verifier-centric architecture has yet touched.
Read essay →