MailSpot

MailSpot is an example that AI can build quality enterprise-level code.

MailSpot is an email program, a software with hundreds of users.

MailSpot was built only using AI coding agents.

MailSpot connects directly to each user's Gmail account.

If you send an email from MailSpot, it is sent from the Gmail servers.

To get that access from Google, the app has to pass Google's cybersecurity tests and compliance audits performed by TAC Security.

These tests are strict, the process is long and it used to cost thousands of dollars.

Before AI, preparing an app to pass took developers weeks or months.

With AI agents, we gathered every security and compliance requirement and built MailSpot to meet them in 2 days.

MailSpot passed the TAC Security audit on the first attempt.

A lot of the features that make MailSpot a great email client have nothing to do with AI, except for that they were built by AI.

But, MailSpot has several AI features.

Every email that arrives is read and categorised by AI.

Emails from customers that require a response get a draft response written.

In order to write good email drafts, MailSpot must have specific knowledge about the user's business and how they have written emails in the past.

Using RAG, the MailSpot system creates and stores a knowledge base for each user, that it uses when it writes emails.

MailSpot is a unified inbox.

Some users have literally connected over 100 email accounts where they can view the emails from all accounts under one inbox.

This feature has nothing to do with AI, but it's a feature that you won't find anywhere else.

A good example of how you can use AI to build something you need that doesn't exist in other services.

Every AI feature in MailSpot is a workflow with a question and an API call to an AI model.

These features are not agents, they are simpler than that.

Essentially, MailSpot is a software program with AI workflows built by AI coding agents.