> ## Documentation Index > Fetch the complete documentation index at: https://docs.antlect.com/llms.txt > Use this file to discover all available pages before exploring further. # Platform architecture > The four layers behind Antlect: the Trusted Data Foundation, the Descriptive Digital Twin, the Semantic Ontology Layer, and the Decision Intelligence Terminal. Antlect is built as four connected layers. Each depends on the one beneath it, and every answer the platform produces passes through all four. The architecture is the reason an Antlect answer carries context, and the reason it can be traced back to evidence. ## Trusted Data Foundation The ground truth. Scientific literature, patents, funding, clinical trials, organisations, and researchers are continuously integrated, validated, and reconciled across more than 100 million data points. The decisive work at this layer is entity resolution. The same researcher appears in a publication record, a patent filing, and a grant award under different name forms, different affiliations, and different identifiers. Resolving those into a single entity is what converts separate datasets into a foundation, and it is why funding, output, and translation can be examined as one system rather than compared as three. The entity types in the graph, and the regions covered in depth. ## Descriptive Digital Twin The structural layer. A continuously evolving representation of the innovation ecosystem: organisations, institutes, laboratories, researchers, capabilities, research infrastructure, intellectual property, and the relationships that connect them. The term descriptive is deliberate. The twin describes the ecosystem as it is and as it changes, rather than simulating what it might do. That restraint is what makes it admissible as evidence: every relationship it holds resolves to a record in the foundation beneath it. The twin becomes visible in [Graph mode](/guides/network-graph), where the entities behind an answer can be examined as a network. ## Semantic Ontology Layer The layer that carries meaning. A policy analyst, a corporate technology scout, and a university researcher will describe one capability in three vocabularies, none of which is wrong. This layer reconciles the taxonomies, classifications, and domain ontologies arriving with the source data into a single conceptual model. Without it, an answer is limited by the questioner's vocabulary. With it, the question reaches the evidence regardless of how that evidence was described. Why reconciling vocabulary changes what an answer contains. ## Decision Intelligence Terminal The layer you work in, where evidence becomes a position. Discovery, knowledge management, AI-assisted analysis, and research briefs operate as one environment rather than as separate tools. How an open strategic question resolves into a structured research brief. ## What the architecture guarantees A question is interpreted through the ontology, resolved against the twin, and answered from the foundation. Each stage is retained, which is why any conclusion can be followed back to the record that produced it. Customer data remains private by design and is never used to train third-party models. See the [Trust Centre](/trust-centre).