Automate invoice data extraction with a cloud agent

Set it up with one message

Invoice Data Extraction reads invoices and returns the fields, and its API is what your agent calls. Everpod gives your agent a real computer that is always on, so it can:

• watch a mailbox for you
• keep permanent records, such as a running spreadsheet
• remember how your business works and how you like extractions handled
• work with files and analyse the results
• do anything else you would do yourself at a computer

Sign up for a free Invoice Data Extraction account, which covers fifty pages a month, create an API key in your dashboard, and send your agent:

I have an account with Invoice Data Extraction (invoicedataextraction.com) and I’ve created an API key for you. Read invoicedataextraction.com/docs/agents and set yourself up to extract data from my invoices with it, then tell me when you’re ready.

The agent asks for the key and tells you where to enter it, then tells you when it is ready.

Send invoices by email

Give the agent a free Gmail account of its own and forward your supplier invoices to it, whatever mailbox they arrive in today. Its password opens the whole mailbox, so an inbox of its own is the right scope. Then tell it when to check:

My supplier invoices arrive at [address]. Check it every morning at 8 (my time zone is [zone]), extract any new invoice PDFs with Invoice Data Extraction, add them to the running purchases spreadsheet, skip anything already done, and tell me here what came in, the totals, and anything that needs my review. Leave every other email alone.

The agent asks for an app password for that account; the Gmail guide has the steps. Ask for a first run straight away rather than waiting for the morning, so you see the whole path once. The scheduling guide covers time zones and changing the schedule later.

Send invoices from your messaging app

Your pod connects to Telegram, Discord, Slack, WhatsApp and more. Attach one PDF or several to a message:

Extract these with Invoice Data Extraction, keep my file names, and send me the results as a spreadsheet.

The rows come back in a few minutes with the spreadsheet attached.

Keep your records in files and spreadsheets

The agent keeps files on its own computer, so it can keep a spreadsheet that grows with every extraction, sort the rows by month or by supplier, keep files in folders, and send you any file when you ask. For a running spreadsheet:

Keep a running purchases spreadsheet of everything you extract, with the same columns as the spreadsheet Invoice Data Extraction returns, and keep every row in the same format. Add new invoices to it, skip anything already in it, and each time tell me what you added, what you skipped and the totals so far.

Connect it to Google Drive and Google Sheets

If you would rather have the spreadsheet in Google Sheets:

1. Ask the agent to connect your ChatGPT subscription, and follow its directions.
2. In your ChatGPT account, add the Google Drive connector and sign in to Google.
3. Tell the agent to put the running spreadsheet in Google Sheets and keep it updated there.

Connect it to your accounting software

If your accounting software has an API, the agent can send the rows straight into it, instead of or as well as a spreadsheet. Tell it which software you use, and it asks you for the access it needs.

When the Invoice Data Extraction API asks a question

You can give your agent context about your business and your extraction tasks. When the Invoice Data Extraction API asks a question about a task, the agent answers it from that context, or brings the question to you in your messaging app to confirm, and you can tell it to remember your answer for next time:

Let Invoice Data Extraction ask questions when a task leaves something unsettled. When it asks, answer from what you know about my business if that settles it, otherwise ask me here to confirm. Remember my answers and use them next time, and tell me in your report what was asked and how it was answered.

What it costs

Extractions your agent runs use the pages on your Invoice Data Extraction account, the same as extractions you run yourself. On the Everpod side, an extraction asked for in chat costs a few cents of model usage and a morning mailbox check around fifteen to twenty cents, out of the usage your pod includes.