{
    "action": "final",
    "content": "**Size:** The dataset has 5 tables (one per month), with 446 rows in total: January 2022 has 115 rows, February 96, March 98, April 95 and May 42. The sizes come from `analytics.table_overview` and the per-month row-count query, and the rendered result shows each month's count.\n\n**Columns:** Each table has the same 13 columns in the raw layer. Nine are business columns: Company Name, Founded, Location, Sector, What it does, Founders, Investor, Amount, Stage. The other four are ingestion metadata: `_ingestion_timestamp`, `_batch_id`, `_source_file`, `_source_system`. The overview metadata reports 9 columns per table, which counts only the business columns. Across 5 tables that is 45 business columns, or 65 columns in total counting metadata. The columns are not combined into one wide table, because the tables share no join key.\n\n**Time period:** The data covers **January to May 2022**, five consecutive monthly snapshots of funding rounds announced in India. The period comes from the table names and the dataset description, not from a date column. The only year-type column, `Founded`, records when each company was founded, and it ranges from 1871 to 2022. It does not describe when the funding happened. `_ingestion_timestamp` is just when the data was loaded.\n\n**Caveats:**\n- May 2022 is the smallest table at 42 rows, and the dataset card notes some of its rows are misaligned or blank.\n- In the May table, `Founded` is stored as text, unlike the other months, where it is numeric."
}