INVOICE PROCESSING

Agent-Built Vendor Dispute Packet From Reconciliation Exceptions

After a statement is reconciled against BigQuery AP data, a CEO agent investigates each exception, drafts a vendor-ready dispute email per discrepancy.

CategoryInvoice Processing
Enginepaperclip
Difficultyadvanced
Triggerevent
Steps5
Setup~25 min

How it runs

The automated pipeline, trigger to output.

  • TriggerReconciliation completes with exceptions
  • ActionQuery full invoice history from BigQueryGoogle BigQueryBigQuery
  • LogicAgent reasons over each discrepancyOpenAI
  • ActionDraft tailored dispute email per exceptionOpenAI
  • OutputFile dispute packet to Dropbox for sign-offDropboxDropbox

What it does

Goes beyond flagging to actually preparing the response. Once a statement is reconciled against AP records in BigQuery, an agent reviews every exception, gathers the relevant invoice context, and composes a clear, professional dispute or clarification email tailored to each discrepancy type (missing invoice, overcharge, duplicate). It assembles the evidence into a packet and files it to Dropbox so an AP manager can review and send.

When to use it

Use this when disputes are frequent enough that drafting them by hand is a real time sink, and you want a human to approve rather than author each message. Best for teams whose AP detail lives in BigQuery.

How it works

  1. 1A completed reconciliation event triggers the flow with the exception set.
  2. 2The agent queries BigQuery for full invoice and payment history behind each exception.
  3. 3It reasons over each discrepancy and decides the appropriate dispute stance.
  4. 4It drafts a tailored vendor email per exception.
  5. 5It compiles the emails plus supporting evidence into a dispute packet.
  6. 6The packet is saved to a Dropbox review folder for AP sign-off.

Set it up

What you configure once, before turning it on.

  1. 1
    Connect BigQueryDatasets, queries, schemas.
  2. 2
    Connect OpenAIModels, embeddings, files.
  3. 3
    Connect DropboxFiles and folders.
  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.