INVOICE PROCESSING

Outlook vendor invoices with PO line-item matching and exception queue

Watches a shared Outlook inbox for vendor invoice attachments, extracts line items, matches them against purchase orders in BigQuery.

CategoryInvoice Processing
Enginesim
Difficultyintermediate
Triggerevent
Steps7
Setup~15 min

How it runs

The automated pipeline, trigger to output.

  • TriggerNew Outlook email with attachment in AP inboxOutlook
  • LogicFilter to messages carrying a PDF or image invoice
  • ActionExtract invoice header and line items with OpenAIOpenAI
  • ActionLook up the referenced PO and its lines in BigQueryGoogle BigQueryBigQuery
  • LogicCompare each line within price and qty tolerance
  • ActionWrite reconciled matches back to BigQueryGoogle BigQueryBigQuery
  • OutputPost mismatched lines to Slack exception queueSlack

What it does

Monitors a dedicated Outlook mailbox (e.g. invoices@) for inbound vendor invoices, pulls the PDF, extracts header and line-item fields, and reconciles each line against the matching purchase order. Clean three-way matches are logged as approved; anything off-tolerance lands in a Slack exception queue with the specific lines flagged.

When to use it

Use it when your AP team manually opens every emailed invoice and eyeballs it against POs in a warehouse. It removes the eyeball step for the 80% that match cleanly and concentrates human attention on real discrepancies.

How it works

  1. 1A new email with an attachment arrives in the monitored Outlook inbox.
  2. 2The flow filters to messages that actually carry a PDF or image invoice.
  3. 3OpenAI extracts structured fields: vendor, invoice number, PO number, and each line's SKU, qty, and unit price.
  4. 4BigQuery is queried for the referenced PO and its line items.
  5. 5Logic compares line by line within a price and quantity tolerance.
  6. 6Matches are written back to BigQuery as reconciled; mismatches post to a Slack exception channel naming the offending lines.

Set it up

What you configure once, before turning it on.

  1. 1
    Connect OutlookMail, calendar, contacts.
  2. 2
    Connect OpenAIModels, embeddings, files.
  3. 3
    Connect BigQueryDatasets, queries, schemas.
  4. 4
    Connect SlackChannels, DMs, threads, mentions.
  5. 5
    Set each agent's modelWe leave models unset so you pick the tier — fast + cheap, or top-quality.
  6. 6
    Tune it to your dataEdit the prompts, filters, and field mappings so it matches how your team works.
  7. 7
    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.