> ## 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. # Decision Intelligence Terminal > How an open strategic question becomes a structured, grounded research brief inside Antlect. Most strategic questions do not arrive well formed. They arrive as a direction of travel: where is our capability thinning, what are we not seeing in this technology, is this position defensible. The difficulty has never been a shortage of information. It is that turning scattered evidence into a position someone will sign off on has historically taken months of manual synthesis, and the result has rarely been auditable once it was written. The reasoning lived in the analyst's head, and the evidence trail dissolved somewhere between the source and the slide. The Decision Intelligence Terminal is the layer where that work happens instead. It is the fourth layer of the [platform architecture](/platform/overview), and the only one you work in directly. ## The question stays open You do not need to know the vocabulary of a field before you can ask about it. Questions are posed in plain language and interpreted through the [Semantic Ontology Layer](/platform/semantic-ontology-layer), which reconciles how the question was asked with how the underlying record was written. This matters more than it first appears. The premise of a question is usually the weakest part of it, because it encodes what you already assume to be true. A question interpreted semantically rather than literally returns the evidence you did not know to ask for, which is the evidence that changes a decision. ## The reasoning stays visible Every question runs as an inspectable trace. Each stage of the analysis is shown as it executes, and the full history is retained, so the route to an answer can be re-examined long after it was produced. Findings are returned as structured evidence rather than prose: the entities that matter, the relationships between them, and the publication, patent, funding, and trial records that substantiate each one. Any statement can be followed back to the record that produced it. Explainability here is a property of the architecture, not a feature bolted onto the output. ## The structure emerges A position rarely arrives fully formed either. Findings worth keeping are retained as artifacts and organised as the picture develops, and a mind map holds the shape of an investigation while it is still forming, branching entities, open questions, and findings as they connect. This is the part of the work that usually happens on a whiteboard and is lost. Here it is retained, editable, and attached to the evidence beneath it. ## The position is written A research brief is where the investigation resolves. It carries the findings, the evidence behind them, and the reasoning that connects the two, in a document you control end to end. The distinction that matters: the brief is authored, not generated. AI assists the drafting, the evidence is assembled for you, and every section remains fully editable. You keep editorial control of what your organisation puts its name to, which is the only arrangement that survives contact with a board, a regulator, or a funding committee. Briefs are written in a full editing environment, carry live charts built from your own findings, and export to PDF for circulation. The practical guide to building, formatting, and exporting a research brief. ## Why this holds up under scrutiny Three properties travel with every output of this layer: * **Grounded.** Findings rest on the [Trusted Data Foundation](/guides/data-coverage), not on a model's recollection. * **Explainable.** The route from question to conclusion is retained and can be re-examined. * **Controlled.** The final document is yours, edited by your people, carrying your judgement. AI is the interface to evidence in Antlect. It is not the source of it.