Discussion
Closed · Answered

For tens of thousands of CSV rows, would you have Buda run a script first?

Leonard Li

Suppose I have a sales CSV with tens of thousands of rows and need deduplication, monthly totals, and products with the biggest changes. I wouldn't need every row repeated in chat.

I'd prefer a script running in the cloud, complete output files saved in Drive, and a chat summary of the findings and checks. How would you ask it to report input and output row counts and identify excluded records?

Report
Kelly
Team
· edited

Prefer a cloud script first: full outputs back to Drive, chat only for findings and check counts. Tens of thousands of rows shouldn’t be replayed in the conversation.

(Diagram: source CSV → cloud script → chat reports counts, exclusions, findings.)

(Local demo: agent session that can run scripts)

Suggested steps

  1. Place the file in Drive and agree fields

Tell Buda the path, dedupe key, and how month columns are read.

  1. Require file outputs

Deduped table, monthly totals, biggest movers — as files, not chat walls.

  1. Mandate reconciliation text

Input rows, output rows, exclusion rules/examples, and output paths.

  1. You review the summary + spot-check files

Open the artifacts when you need detail.

How to confirm

  • Input − excluded ≈ primary output (with any remainder explained).
  • Spot-check excluded rows against the stated rules.
  • Full results open in Drive; chat isn’t flooded with raw rows.

Boundaries

  • Reading “in chat” is fine for a sample insight; auditable dedupe/totals belong in scripts.
  • Encode the script brief in a Skill for the next similar CSV.
  • Keep result-file permissions aligned with the source when data is sensitive.
Report

Your reply

Sign in to join the discussion. Reading is always open; you only need an account to post.

Sign in