What is a Halcyon Query
A Query is a natural-language question or prompt that generates a cited response from Halcyon's data catalog. Queries return higher-quality responses when Halcyon's data catalog is filtered to relevant sources.
How the AI works, under the hood
When you run a query, Halcyon follows a two-step process to produce a well-researched response:
- Retrieve. Halcyon searches your filtered documents for the most relevant information — including text, images, tables, and charts — related to your question.
- Generate. AI extracts that relevant information and uses it to generate a response, with citations pointing back to the original source excerpts.
This is why filtering matters so much: the better you narrow the source set in step one, the more accurate and useful the generated answer in step two. Halcyon's software is also purpose-built for energy-industry language and terminology — including acronym disambiguation — so responses are more accurate and relevant than a general-purpose tool reading the same filings.
Reading your query response
Once a response is generated, it's organized so you can read the answer and verify it against the source material. There are three parts:
Analysis — the AI-generated written answer to your prompt. Throughout the text you'll see inline citations; click any citation to jump to the exact source excerpt it's based on, so you can confirm the claim in the original filing.
Documents — the subset of documents Halcyon used to construct the response. The citations allow you to see exactly which filings informed the answer, open them in full, or dig deeper into a source that looks relevant.
Figures — any tables, charts, or visual elements from the source documents that informed the response. When a number or trend in your answer comes from a table or chart, this is where you'll find the original visual.
Got questions or feedback about this page? support@halcyon.io
Powered by