From search to structured dataset in three steps

Right now, the Halcyon team will build the grid and add the rows (search constraints) with some preliminary columns. From there, add your questions as columns, spot-check the answers, then scale up. This guide walks through each step using a real example — a grid of large generator interconnection agreements (LGIAs) filed in Texas.

Each column needs four things:

For the Texas LGIA grid, the columns were:

When you save a column, Halcyon runs the query against every row's documents and the cells fill themselves in. Larger grids take a little longer to populate.

Before you grow the grid, click into a handful of cells and read the citations. Every answer is traceable: the cell detail shows the exact query that ran, plus source citations pointing back to the original documents. This is the trust layer — you can verify any value against the original filing.

Use what you find to refine your column queries, then apply them to the full dataset. A good rule: experiment with search filters and column queries on a handful of rows first, and check that they hold up across different time periods and document types.

If a cell shows a purple checkmark, the value has been human-verified by the Halcyon team. Right now, only the Halcyon team can verify cells.

Grids export to Excel for downstream analysis. One important caveat: exports include the values, but not the cell-level citations — if you or a colleague needs to verify sources, use the live grid.

☢️ [IMPORTANT] ☢️ Right now (as of August 20, 2026), Grids exist at the account level, but any changes made are user-specific. This means that if you add a column to a Grid in your account, your teammate will not be able to see that column in their version of the Grid. We’re working on a way to make sharing Grids easier within and across teams, and if you’d like to share a version of a Grid you made with a colleague, reach out to Halcyon directly: support@halcyon.io