One platform.
Four ways to start.

No. Customer recordings, documents, and transcripts are never used for model training. Ever.

Every allowance is a hard cap. New work that uses that allowance pauses until it resets or the plan is upgraded; existing transcripts, answers, artifacts, and verified visual evidence remain available.

No. Transcript-first processing creates the recap, transcript, transcript-derived artifacts, and transcript Q&A without using the visual-evidence allowance. You choose if and when to analyze scenes, on-screen text, and frame evidence.

There is no monthly audio or transcript-processing hour allowance. You can keep transcribing while your workspace remains within its published recording-library, retention, per-file size, and other plan limits. Optional visual evidence has its own separate monthly allowance.

They stay in your workspace alongside the structured evidence, and count against your plan storage. Deleting a recording removes both the media and everything derived from it.

PDFs use their extracted page count. Word files use content-based page-equivalents because DOCX has no reliable page boundaries. Markdown uses one page-equivalent per 3,000 characters, rounded up. Billing shows the resulting monthly usage.

A sub-allowance inside your document pages, not a second pool. Reading a PDF page optically - so its text, tables, and positions can be cited precisely - is by far the most expensive step, so it gets its own ceiling. Each PDF page consumes one document page and one PDF (OCR) page at the same time, which means PDFs are capped by the smaller number, while Word and Markdown never touch the OCR allowance and can spend the rest of the document-page budget.

Yes. Evidence Q&A questions are one monthly pool: a question asked of a recording, a document, or a mixed project scope draws from the same allowance. Deleting a document frees its stored-document slot immediately.