Why is email treated different than any other data?

I don’t know exactly how Gmail stores mail on the backend. I do know Microsoft 365 runs on Exchange, and I’ve been an Exchange admin for a while. Exchange sits on top of ESE, a custom B-tree database engine Microsoft has been hauling around since the 90s. Working with it at scale makes no sense to me. Email is text. Put it in Postgres like any other text and move on.

The only thing actually special about email is the delivery protocols. Before APIs and REST, the world settled on SMTP, POP, and IMAP, and we’ve been writing wrappers around those wrappers ever since. Stick a connector in front of a normal database, inject messages as data, process in a workflow, and the whole game stops being “extract from a giant email provider” and starts being a direct data pipeline you control.

AI likes text. So strip email down to text the second it lands. Save attachments in blob storage and link them. Link the email to a customer automatically. Email, docs, ideas, meeting notes, work product, all stored as checkpoints in a timeline of knowledge for that client.

When the data is needed, the move is to combine information and produce an artifact, as Anthropic might say, rather than open a document.

The theme is that information should live at the lowest level it can. Raw text, tagged and linked, easy for an AI to use and easy to manage. When you need to send a doc, produce a contract, or look back at a meeting, you export to Word, PDF, or PowerPoint on the way out. You stop opening five revisions of a doc from a year ago trying to remember which one had the good paragraph.

A side benefit of ingesting email as text first is security. There is no executable to run. No attachment lurking in a preview pane. Phishing gets harder because you are not relying on someone else’s identity stack to render the message.

Once ideas are artifacts in seconds and text snippets first, the middle layers stop earning their keep. The proprietary stores, the connectors stacked on connectors, the UI built to compensate for both. The cost of email was always the mailbox.