Aletheia for Investigations & Intelligence Triage
Investigations documents and intelligence reports go in. Entities, relationships and facts come out. The model proposes. The platform verifies. The analyst decides.
Documents and Reports go in. Entities, relationships and facts come out. The model proposes. The platform verifies. The analyst decides.
Information overload across intelligence, law enforcement, compliance and risk domains. Fragmented documents and reports pile up in queues with no consequence-based prioritisation, no cross-referencing, and no audit trail. Reports arrive faster than they can be read, triaged, or acted upon.
The ontology is not designed in advance. It is produced from the work. Aletheia inverts the ontology-first orthodoxy. Types, attributes and relationships emerge from what the reporting actually contains, not from a steering committee's prediction. The graph grows to fit the mission, not the other way around.
Entities, relationships and typed facts drawn from text. Every proposal is bound to the sentence in the source that produced it.
Multiple independent signals are cross-referenced. Evidence is required. Low-confidence proposals are held back. Confidence is measured, not assumed.
Confirms, overrides, merges, splits. Final resolution is the analyst's call, not the model's. The ontology becomes a living artefact of the centre's own work.
A transparent, citation-backed data processing pipeline with five stages: Ingest, Extract, Resolve, Analyse, Produce. Every extracted fact links to the chunk that produced it. Every write is recorded in an immutable ledger.
Documents and reports are ingested from multiple sources — files, feeds, partner systems.
Entities, relationships and typed facts are extracted from text and bound to their source.
Entities are resolved and merged across sources. Confidence is measured, not assumed.
Graph algorithms and analytical functions surface patterns, centrality, paths and risk.
Dossiers, profiles and briefings are produced with full citation and provenance.
Consequence-based prioritisation. The system triages by risk, not by arrival order.
Connections across sources. The graph surfaces relationships a human reader would miss.
Time-aware analysis. Temporal traversal and time-travel queries reconstruct what the system held to be true on any date.
Audit and provenance first-class. Every fact links to its source. Every write is recorded.
Provenance and time-travel are not features bolted to the top — they cut through every layer, from source to output. Audit is the architecture, not a finishing step.
Dossiers, profiles, briefings.
Resolution, override, query.
Risk, pattern, tempo, disclosure.
Ingest, extract, resolve, analyse.
Documents, reports, feeds, files.
A vector database with graph features bolted on cannot answer multi-hop questions. Symplast is a graph at its core — entities, relationships and typed facts, queryable across the chains real intelligence work requires.
Every extracted fact links to the chunk that produced it. Every write is recorded in an immutable ledger. Ask what the system held to be true on any historical date and the answer is reconstructable, not approximated.
The model proposes. The scaffolding verifies. The operator decides — overriding resolution, exporting findings, keeping the wheel. The analyst is the source-of-truth on the system, not the other way around.
The platform ships with the breadth required to make the graph operational from day one.
Community detection, PageRank centrality, betweenness centrality, closeness centrality, multi-hop pathfinding, link prediction, network topology, temporal traversal, time-travel queries, conversational query, entity resolution, confidence weighting.
Consequence-based prioritisation replaces FIFO processing. Every input is weighted, cross-referenced and routed by risk.
Fragmented folders become a connected graph. Relationships across sources surface in a single view.
Multiple collection streams fuse into a single corpus with gaps highlighted and provenance preserved.
Continuous risk assessment with threshold alerts. Noisy incident data refined into a clear root-cause chain.
Aletheia runs on Symplast, inside the fusion centre. On your infrastructure, under your control, behind your accreditation boundary.
Air-gapped, on-premises, or sovereign cloud at the customer's discretion. No tenancy in foreign clouds. The data is yours. The decisions are yours.
Integrates with existing case management, records and partner-agency intelligence systems through standards-aligned data exchange. Role-based access and disclosure controls are first-class concepts.
Because the ontology is produced from the data, there is no separate ontology phase to fund and wait through. We deploy forward, work alongside your operators, prove the capability on your data.
Deployment modes: Air-gapped, On-premises, Sovereign cloud.
Aletheia on Symplast is sovereign Australian capability. The code, the data and the decisions stay onshore — under Australian law, under Australian accountability. Built in Australia by Australian-owned companies whose founders come from government, law enforcement and national security backgrounds.
Australian-owned. Australian-built. Informed by lived experience.
Symplast is patent-pending technology of Psithur Holdings Pty Ltd.
People Tools Instructions Pty Ltd. Request a briefing to learn how Aletheia on Symplast can be deployed in your fusion centre.
Investigations documents and intelligence reports go in. Entities, relationships and facts come out. The model proposes. The platform verifies. The analyst decides.