INVOICE PROCESSING

Emailed Invoice PDF Drift Detector

Parses vendor invoice PDFs arriving in Gmail, extracts the billed amount with an LLM, checks it against the contracted recurring price in Postgres.

CategoryInvoice Processing
Enginesim
Difficultyintermediate
Triggerevent
Steps5
Setup~15 min

How it runs

The automated pipeline, trigger to output.

  • TriggerNew invoice email in Gmail labelGmailGmail
  • ActionExtract amount and vendor from PDFOpenAI
  • ActionFetch contracted rate from PostgresPostgreSQLPostgres
  • LogicClassify match / minor / major drift
  • OutputFlag drift to Microsoft TeamsMicrosoft Teams

What it does

Many vendors still bill by emailing a PDF. This workflow reads those invoices the moment they land in a Gmail label, uses an LLM to pull the vendor name, period, and total, then matches that vendor to its contracted recurring amount in your Postgres ledger. Any invoice that deviates beyond the allowed band gets routed to a Microsoft Teams channel so the right person reviews it before it's paid.

When to use it

Use it for vendors who don't bill through a payment platform and instead send invoice attachments. It removes the manual step of opening every PDF and remembering what the contracted rate was supposed to be.

How it works

  1. 1A new email with a PDF attachment lands in the watched Gmail label.
  2. 2An OpenAI step extracts vendor, billing period, and total amount from the PDF text.
  3. 3The flow queries Postgres for that vendor's contracted recurring amount.
  4. 4A logic step compares billed vs. contracted and classifies it as match, minor drift, or major drift.
  5. 5Matches log silently; drifts post to Microsoft Teams with the extracted figures and source email link.

Set it up

What you configure once, before turning it on.

  1. 1
    Connect GmailRead, draft, send, label.
  2. 2
    Connect OpenAIModels, embeddings, files.
  3. 3
    Connect PostgresAny Postgres URL — query, write, migrate.
  4. 4
    Connect Microsoft TeamsChannels, chats, files.
  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.