DOCUMENT OPS

Scanned receipt batch to BigQuery expense ledger

Watches a Dropbox folder for new scanned receipts, extracts vendor, date, total, and tax with a vision model, and appends each parsed receipt as a structured row in BigQuery.

CategoryDocument Ops
Enginesim
Difficultyintermediate
Triggerevent
Steps5
Setup~15 min

How it runs

The automated pipeline, trigger to output.

  • TriggerNew file added to Dropbox receipts folderDropboxDropbox
  • ActionDownload receipt file from DropboxDropboxDropbox
  • ActionExtract fields with Hugging Face vision modelHugging FaceHugging Face
  • LogicValidate total parsed; skip and flag if unreadable
  • OutputInsert structured row into BigQuery expenses tableGoogle BigQueryBigQuery

What it does

Turns a drop folder of phone-scanned and emailed receipts into clean, queryable expense rows. Each new image or PDF in Dropbox is read by a Hugging Face vision-language model that pulls vendor name, transaction date, line-item subtotal, tax, and grand total, then writes one normalized record to a BigQuery expenses table.

When to use it

Use it when your team dumps receipts into a shared Dropbox folder and finance needs them as analyzable data, not a pile of JPEGs. Ideal for monthly close, reimbursement prep, or feeding a spend dashboard without manual data entry.

How it works

  1. 1A new file landing in the watched Dropbox receipts folder triggers the run.
  2. 2The file is downloaded and its bytes passed to a Hugging Face OCR/vision model that returns vendor, date, currency, subtotal, tax, and total as JSON.
  3. 3A logic step validates the extraction: if the total is missing or unparseable, the receipt is skipped and flagged.
  4. 4The cleaned record, with the original file path retained, is inserted into the BigQuery expenses table for reporting.

Set it up

What you configure once, before turning it on.

  1. 1
    Connect DropboxFiles and folders.
  2. 2
    Connect Hugging FaceModels, datasets, spaces — the open-source hub.
  3. 3
    Connect BigQueryDatasets, queries, schemas.
  4. 4
    Set each agent's modelWe leave models unset so you pick the tier — fast + cheap, or top-quality.
  5. 5
    Tune it to your dataEdit the prompts, filters, and field mappings so it matches how your team works.
  6. 6
    Test, then turn it onRun once against a sample, confirm the output, then enable the trigger.

Run this workflow in your colony.

14-day trial. No DevOps. No Sales call. Provisioned in under a minute.