Pcloudy/powered by Arc AI

Documents

Product specs, test plans and compliance circulars become queryable evidence for every failure.

Upload a document

PDF, Word, Excel, Markdown, text, CSV, HTML · max 12 MB

Drop a file here, or browse

Arc reads the text, splits it at headings into ~400-word passages with overlap, embeds each one and links the document to the feature it talks about.

Test case, defect and requirement lists

An Excel, JSON, CSV or TSV export is read row by row, so every test case becomes its own record linked to a product area. Run this once for files uploaded earlier.

Re-read documents with the improved splitter

Documents read earlier were cut into smaller pieces than intended, which weakens search. This reads them again and rebuilds their passages.

Tidy section titles

Some passages carry a table's column headers ("ID SCENARIO TYPE PRIORITY REF") or a stray fragment as their section title, which is what citations show. This corrects the titles; the text and the search index are left alone.

Search the index

Search accuracy

Save questions your teams actually ask, with the document or wording the right answer should contain. Running the set replays every question through the real search and reports where the right passage landed — so a change to search is measured, not assumed.

Sign in to run the check.

Answer accuracy

Are Ask Arc's answers right, and backed?

Save questions your teams really ask, with the wording a correct answer must contain and the record it must cite. Running the set replays each one through the real answer path.

Sign in to save and run answer checks.

Indexed documents

0 indexed · 0 passages · 0 embedded

FileKindSizePassagesFeatureAddedStatus
Loading…