Data teams keep rediscovering the same need: a corpus of video transcripts — for analysis, for training material, for the internal search index, for whatever the quarter's project is. And they keep meeting the same cost anxiety: transcript acquisition at volume feels like a metered faucet nobody wants to be the one who left running.

The veridive cost structure has a shape worth designing around: imports are the paid step; transcript pulls from imported videos are free. That asymmetry is a pipeline design — pay deliberately once at the import gate, then read freely forever. Add a meter check at the gate and the faucet anxiety becomes a budget line.

library_import_youtube_video is the paid gate: the corpus enters the library deliberately, batch by batch, tagged for the project. get_youtube_transcript is the free read: timestamped transcripts pulled per video, batchable across the whole imported set, re-pullable whenever the downstream format changes — at zero marginal cost.

get_credit_status keeps the gate intentional: check the meter before each import batch, size the batch to the budget, and the pipeline's spend is a decision made at one point instead of a surprise distributed across a thousand calls. The output: a tagged transcript corpus, acquired at a known cost, readable for free from then on.