◳
Loading...
Loading...
Product Overview
The TACITUS platform combines deterministic graph structure with probabilistic AI reasoning — creating conflict intelligence that is grounded, auditable, and actionable.
Increase accuracy for human friction professionals. Structure what LLMs cannot.
The Foundation
Core Engine
In Active DevelopmentEvery TACITUS product is powered by the same foundational engine: a proprietary 8-primitive Conflict Grammar encoded as a typed knowledge graph. The engine combines deterministic structure (the graph never hallucinates) with probabilistic reasoning (LLMs extract and reason) — creating a neurosymbolic system that thinks about conflict the way trained professionals do.
Our simplified open-source pipeline shows the extraction and structuring stages. The full Core Engine adds multi-tenant storage, the complete Agentic Conflict Ontology, enterprise provenance, and graph-grounded LLM reasoning.
8 Conflict Primitives — Conflict Grammar Ontology
Actor · Claim · Interest · Constraint · Leverage · Commitment · Event · Narrative
Application Suite
Each product uses the same conflict graph — so your data works across your entire workflow.
All products are experimental and under active development. Demo applications represent early-stage prototypes and are provided for evaluation purposes. Features, interfaces, and capabilities are subject to change.
The Conflict Operating System
The comprehensive TACITUS workspace. Ingest documents, build conflict knowledge graphs, track actors and claims over time, manage multi-party cases, and access AI-grounded analysis — all in one platform.
AI-Assisted Conflict Navigation
The accessible entry point to TACITUS. Structure a dispute quickly — identify actors, map claims, surface hidden interests, and explore resolution pathways with AI assistance.
Conflict Scenario Simulation
Test before you deploy. Simulate conflict trajectories, stress-test negotiation strategies, and explore 'what-if' scenarios against structured knowledge graphs.
Deep Conflict Intelligence
For professional mediators and analysts managing complex, multi-party disputes. Cross-case pattern recognition, deep temporal analysis, and AI-grounded briefing generation.
Architecture
emails, docs, transcripts
Core Engine
Pipeline
A 4-stage pipeline where AI does what it's good at — and the graph does the rest.
Raw sources: emails, transcripts, docs, notes.
LLMs identify actors, claims, events, evidence. Extraction only, not judgment.
Agentic Conflict Ontology: type binding, relation mapping, temporal ordering, provenance linking.
Graph-native storage. API-queryable. Deterministic. LLM-groundable.
The Problem
LLMs flatten time into vectors. 'Promise made' and 'promise broken' become statistical neighbors, not a causal sequence spanning months.
Transformers learn correlations, not causation. 'Before' doesn't mean 'caused.' This distinction is fatal in any dispute context.
Generative models cannot trace outputs to inputs. In legal, HR, and diplomatic contexts, unverifiable claims are worthless.
All demos are experimental early-access prototypes. Try them and tell us what you think.