Loading TACITUS
Loading TACITUS
Ontology Augmented Generation · Neurosymbolic Architecture
Dialecticais the TACITUS core engine — the neurosymbolic context backend for conflict and policy intelligence. Standard AI treats disputes, reports, and policy situations as text to summarize. Dialectica treats them as structured systems to reason about — typed knowledge graphs that make context legible to machines and auditable by the professionals who act on it.
Named after the ancient art of reasoned dialogue, Dialectica goes beyond retrieval (RAG) and entity linking (GraphRAG) to provide Ontology Augmented Generation— a conflict-specific ontology grounded in 30+ theoretical frameworks that preserves temporality, causality, and provenance. The engine that powers every TACITUS product.
Every event timestamped. Every claim version-stamped. The graph reconstructs what was said, when, in what sequence.
Not "A then B" but "A caused B via mechanism M." The difference between a log and intelligence.
Every assertion traces to a source document, timestamp, and actor. A full audit trail by design.
The Real Problem
The hardest questions in conflict resolution are structural, temporal, and relational. They require connecting versioned claims across time — not retrieving similar text chunks.
“Did the manager’s behavior escalate after the complaint was filed, or was the pattern already present?”
“Which commitments did Party B acknowledge in writing that they later denied agreeing to?”
“Which parties shifted position on water rights after March, and did new coalitions form?”
“Beneath the property arguments, what does each sibling actually need — and where is there overlap?”
The Agentic Conflict Ontology
The Agentic Conflict Ontology — the kernel grammar that DIALECTICA reads and writes — has moved to its canonical home in research. Eight primitives that do not change, dynamic per-case extensions that do, MIT-licensed and forkable.
How It Works
Load documents, transcripts, emails, case notes. LLMs identify actors, claims, events, evidence. Extraction only — the LLM is a pattern recognizer, and that’s exactly what we use it for.
The Agentic Conflict Ontology applies type binding, relation mapping, temporal ordering, and provenance linking. Every node typed against the 8-primitive Conflict Grammar. This is where structure becomes intelligence.
Graph-native storage. API-queryable. Deterministic on keyed facts. LLM-groundable on open questions. The graph never hallucinates. The LLM never flies blind.
The Open Engine
The engine is not a monolith. It is composed of AGON — the Rust conflict-vision layer that reads who, what, and why — and KAIROS — the Rust temporal-vision layer that reads when. Both are still open source. DIALECTICA binds them against the ACO kernel and produces the typed graph PRAXIS reasons over.
Temporal vision · Rust
Turns long-form political, legal, and diplomatic text into a temporal knowledge graph: dated events, canonical actors, commitments, episodes, Allen-13 temporal relations, and source-grounded evidence spans. The "when" of the engine. Rust-first, deterministic core, embedded Axum workbench.
Conflict vision · Rust
Reads messy human conflict text and produces typed, evidence-backed primitives: actors, claims, denials, commitments, contradictions, escalation signals, and friction matrices. Preserves disagreement as data instead of smoothing it into prose. The "who, what, and why" of the engine.
Composer · Neurosymbolic engine
The composer. Takes AGON’s conflict primitives and KAIROS’s temporal scaffolding, binds them against the ACO kernel grammar, and produces the typed, four-layer graph (GND, CTX, EVD, RZN) that PRAXIS, Wind Tunnel, and CONCORDIA reason over. Ontology-Augmented Generation. Provenance-bound. Contestable.
Why split it
Conflict structure and temporal structure are different problems. A single end-to-end model collapses them and loses both. KAIROS gets time right. AGON gets disagreement right. DIALECTICA composes them so the graph that lands in PRAXIS is provably built on each.
DIALECTICA · neurosymbolic composition · live
AGON parses conflict structure. KAIROS parses time. DIALECTICA binds both against the ACO kernel and writes them into four reasoning layers — ground, context, evidence, reasoning — every node carrying its source provenance.
Sources
docs · cables · transcripts
AGON
conflict vision · Rust
who / what / why
KAIROS
temporal vision · Rust
when
PRAXIS
workbench
GND · Ground
Typed facts. Actor/Claim/Event nodes bound to source spans.
CTX · Context
Institutional roles, frameworks, prior-case inheritance.
EVD · Evidence
Verified spans, confidence, contradicting attestations.
RZN · Reasoning
Inferences, hypotheses, contestability annotations.
DIALECTICA · four reasoning layers · 3D
The composed graph carries four reasoning layers — ground, context, evidence, reasoning. Drag the stack to rotate it. Click a layer to inspect what lives there. Every node on every layer points back to the layer below.
GND · Ground
Typed facts · source-boundThe base layer. Every Actor, Claim, Event, Commitment, Interest, Constraint, and Date is a typed node bound to a precise source span (doc · paragraph · offset). The model cannot make a claim that does not exist here.
doc-042 ¶1doc-042 ¶1doc-042 ¶2doc-042 ¶3doc-044 ¶4doc-044 ¶5Layer keys ▸
Every node on a higher layer carries an explicit reference to one or more nodes on the layers beneath it. That is what makes the graph contestable: an analyst can climb down from reasoning to evidence to context to ground, and challenge or replace anything along the path.
0+
dates
0+
events
0+
episodes
0+
Allen-13 relations
From a single KAIROS run on the Meridian Compact crisis dossier.
KAIROS · Allen-13 interactive
What it says
A begins first, ends inside B.
Drag the bars to move them; drag the edges to resize. Allen-13 is the kernel of time-aware reasoning — KAIROS computes these between every pair of episodes in a case file.
AGON · friction matrix
| Nordalia | DRM | UN Obs. | Estral. | Mediator | |
|---|---|---|---|---|---|
| Nordalia | |||||
| DRM | |||||
| UN Obs. | |||||
| Estral. | |||||
| Mediator |
Selected friction · intensity 0.84
Ceasefire breach contested
Claim
Nordalia: DRM violated the 5-km halt on 13 March.
Counter
DRM: movement was administrative reposition, not offensive.
Source
doc-042 ¶3 · doc-044 ¶1AGON outputs this matrix for every case it reads. The numbers above are illustrative, but the shape — actor × actor × evidence — is the actual data structure.
The Agentic Conflict Ontology
Not generic NLP categories — the reasoning patterns of trained conflict professionals, grounded in Fisher/Ury, Galtung, Glasl, Lederach, and 30+ theoretical frameworks.
Principal, Agent, VetoPlayer, BridgeActor, Mediator — typed roles, not just “person mentioned”
Position, Interest, RedLine, BATNA — what actors assert, need, and won’t concede
The need beneath the position. Per Fisher/Ury: interests unlock resolution
Legal, institutional, temporal limits that bound the solution space
Asymmetric power — who can compel action and through what mechanism
Promises and obligations tracked over time with version-stamped provenance
Timestamped occurrences with typed causal links — the atoms of the timeline
Competing stories about the same events. Divergence reveals where mediation begins.
Typed node
Each extraction becomes a primitive with a role, confidence, timestamp, and source span.
Typed edge
Relations such as contradicts, commits-to, constrained-by, escalated-from, and aligns-with are explicit.
Graph layer
GND, CTX, EVD, and RZN keep raw facts, context, evidence, and reasoning separate.
Contestability
Analysts can inspect, revise, reject, or annotate the graph before generation uses it.
Dialectica · Work in Progress
We're building in public from NYC. Explore the Dialectica engine, inspect the open-source pipelines, or read the research.
Frequently asked
What the neurosymbolic engine is, how it composes AGON + KAIROS, and how it differs from RAG / GraphRAG.