Ask it anything, against all of itIt works your question across every record on file, then answers
Press the button on a rule, a competitor, a buying office, a budget line, a code, a vehicle, an award or an open notice, and you get a short reading of it: what it means, who it hits, what it is worth, what to do.
Ask a question in plain words and it goes to work. It breaks the question into its parts, shows you the plan, and searches the award history, the open notices, the subawards, the assistance awards, the vehicles, the buying offices, the resolved vendors, the contracting people, the sole-source justifications, the budget exhibits, the labor rates, the exclusions, the rules and the bills for each part, up to six searches a question. Millions of records, every fiscal year back to FY2001, not one page of one table.
Then it answers, and every figure in the answer is counted over the whole matching set rather than over a sample. It tells you which searches it ran and how much each one found, so you can open any of them and check the number yourself. It works only from what those searches returned: where the record does not settle a point it is required to say so rather than fill the gap, and every AI block is labelled, names the model that wrote it, and says what it was written from.
- A reading on every record screen, not one chat box bolted to the side
- Work types, processes and materials on every new notice, so your filters work without keywords
- One question can open up to six searches across the whole record, and it names every one it opened