Problem
The company received a steady flow of emails of different kinds. Someone had to open each one, decide what it was, then answer it, forward it to the right person or file it. Most of these decisions repeated from day to day, yet they still took a person’s time.
Solution
- An AI agent connected to the company’s Google mailbox through the Gmail API.
- For each new email the agent looks at the sender, the subject, keywords and other patterns in the message, and uses a language model to work out what the email is about.
- Based on that, it carries out the actions set up for that type of email: an automatic reply, forwarding to a colleague, labels and marks, or moving the email to the right folder.
- The actions are configured as rules, so the team can add a new type of email or change what happens to it without rebuilding the agent.
Result
- About 70% of incoming emails are handled automatically: answered, forwarded, labelled or filed as they arrive.
- About 1 hour a day of manual inbox work saved.
- The team spends its inbox time on the emails that need a person.