Loading TACITUS
Loading TACITUS
BUILD ON TACITUS · UNDER ACTIVE DEVELOPMENT
Three ways to put Dialectica inside your own application: managed API, embedded platform SDK, or the open-source pipeline. Add graph context, provenance, and ontology-aware reasoning to products for mediation, policy analysis, legal work, risk, and institutional knowledge.
THREE PATHS
MANAGED
Hosted conflict and policy intelligence. Send text, receive a typed graph. Python and TypeScript clients are in private beta; request a key for pilot access.
Request API accessPLATFORM
Embed TACITUS inside your own product. Use the ontology, extraction, policy/context graph, and query layers directly. Enterprise pilot scope is defined with the team.
Pilot conversationOPEN SOURCE
Clone, self-host, extend. MIT-licensed reference implementations of the ontology, the extraction pipeline, and the engine.
GitHub ↗QUICK START
from tacitus import Client
client = Client(api_key="tk_live_***")
graph = client.extract(
text=open("grievance.txt").read(),
domain="human_friction",
)
for actor in graph.actors:
print(actor.name, "->", [c.text for c in actor.claims])
for commitment in graph.commitments:
print(commitment.source_span, commitment.status)
import { Client } from "@tacitus/client";
const client = new Client({ apiKey: process.env.TACITUS_API_KEY! });
const graph = await client.extract({
text: await readFile("grievance.txt", "utf8"),
domain: "human_friction",
});
for (const actor of graph.actors) {
console.log(actor.name, actor.claims.map((c) => c.text));
}
curl -X POST https://api.tacitus.me/v1/extract \
-H "Authorization: Bearer $TACITUS_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"text": "The Government of Nordalia announced ...",
"domain": "armed_conflict"
}'Public endpoint details will be finalized during the managed API beta.
Add structured conflict memory to a live mediation tool. Session-level commitment tracking built in.
Turn policy corpora into queryable actor/claim/commitment/constraint graphs. Cite every finding back to a source.
Ground a negotiation copilot in a typed ontology instead of in a context window. Fewer hallucinations.
Structure grievance intake, flag contradictions, separate claim from commitment automatically.
Feed proposed policy into Wind Tunnel via the SDK. Observe behavioral-tribe responses programmatically.
Build PRAXIS-class document understanding on your own corpus — typed graph, citations, and graph-grounded answers, vendor-agnostic.
THE ONTOLOGY UNDERNEATH
OPEN SOURCE
Rust-first temporal vision layer. Turns long-form political, legal, and diplomatic text into a temporal knowledge graph — dated events, canonical actors, commitments, Allen-13 temporal relations, episodes, and source-grounded evidence spans. The "when" layer of the engine.
Rust conflict-intelligence engine. Reads messy human conflict text and produces typed, evidence-backed primitives: actors, claims, denials, commitments, contradictions, escalation signals, and quality gates. Preserves disagreement as data. The "who/what/why" layer.
The reference pipeline: ingest, extract, structure, query. MIT-licensed.
Reference implementation of Dialectica engine components.
The Agentic Conflict Ontology. Pydantic + OWL/Turtle dual export.