AI-Powered Email Data Extraction for Agents

Names, email addresses, phone numbers, dates, dollar amounts, order numbers, tracking numbers, action items, locations, deadlines. You define the schema, and the extraction returns JSON matching it.

Yes. LobsterMail extracts text from PDF attachments and includes it in the extraction pipeline. The agent can pull structured data from both the email body and attached documents.

You define a schema with the fields you want — types like string, number, date, array. LobsterMail's AI-powered extraction (built on Claude) parses the email and returns clean JSON matching your schema.

JSON. You define the field names and types in your schema, and the extraction returns a typed object. Ready to insert into a database or pass to another service.

Yes. The extraction handles both plain text and HTML emails. It reads through the rendered content regardless of the email's formatting.

The AI-powered extraction adapts to different formats. An invoice from Stripe looks nothing like one from your hosting provider, but the extraction pulls the same structured fields from both.

The extraction returns JSON. What you do with it is up to you — insert into Postgres, push to a webhook, write to a spreadsheet, feed it to another agent. LobsterMail handles the parsing.

Not automatically. You call markRead() when you're done processing. This gives you control — if extraction fails or your database insert fails, the email stays unread for retry.

It's powered by Claude, so it handles messy real-world emails well. For structured fields like dollar amounts and dates, accuracy is high. For ambiguous free-text fields, results depend on the email content.

Yes. You pass a schema object to extractStructured() with whatever fields you need. Order numbers, line items, shipping dates, customer names — define it and the extraction fills it.

Yes. The extraction reads the full email content, including forwarded message bodies. If the data is in the email, the extraction can find it.

Claude handles multilingual content natively. The extraction works on emails in English, Spanish, French, German, Japanese, and most other languages without any configuration.