> ## Documentation Index > Fetch the complete documentation index at: https://docs.allgoodhq.com/llms.txt > Use this file to discover all available pages before exploring further. # Overview > Mary reads, classifies, and acts on every reply that lands in your marketing inbox — automatically. When you send marketing emails, replies come back. Lots of them. Unsubscribes, out-of-office messages, people who've changed jobs, bounce notifications, genuine sales inquiries — and a mountain of spam. Sorting through all of this manually is tedious and error-prone. **Mary's Email Reply Management (ERM)** handles this automatically. Every reply that arrives in your configured inbox is read and classified by Mary's AI. Based on the category it falls into, Mary can extract key pieces of information from the email and then trigger downstream actions — like updating a Marketo record, adding someone to a static list, or creating a new contact. ## The three building blocks The whole system is built around three concepts you configure yourself: Define the buckets replies fall into. Give each one a name and a plain-English prompt describing how Mary should recognize it. Tell Mary which fields to pull from emails in a given category — replacement contacts, return dates, new email addresses, and more. Configure what happens automatically after classification — syncing Marketo fields, adding to static lists, and more. ## How an email gets processed When a reply hits your inbox, here's exactly what happens: The reply lands in your dedicated allGood inbox address (e.g., `8IEJ8ZOP@parse.allgoodhq.app`). Every message sent to this address is automatically picked up and queued for processing. Mary reads the full email — subject line, body, and metadata — and compares it against the categories you've defined. She picks the best match based on the description you wrote for each category. Mary reads the full email — subject line, body, and metadata — and compares it against the categories you've defined. She picks the best match based on the description you wrote for each category. If you've set up data extraction for that category, Mary pulls specific pieces of information out of the email — for example, a replacement contact's name and email address from a "Left Company" auto-reply. Mary reads the full email — subject line, body, and metadata — and compares it against the [categories](/use-cases/erm/categories) you've defined. She picks the best match based on the description you wrote for each category. If you've set up data extraction for that category, Mary pulls specific pieces of information out of the email — for example, a replacement contact's name and email address from a "Left Company" auto-reply. With the email classified and data in hand, Mary fires the actions you've configured for that category — updating a Marketo lead, adding someone to a static list, and so on. If you've set up [data extraction](/use-cases/erm/data-extraction) for that category, Mary pulls specific pieces of information out of the email — for example, a replacement contact's name and email from a "Left Company" auto-reply. With the email classified and data in hand, Mary fires the actions you've configured for that category — updating a Marketo lead, adding someone to a static list, and so on. With the email classified and data in hand, Mary fires the [actions](/use-cases/erm/actions) you've configured for that category — updating a Marketo lead, adding someone to a static list, and so on. ## Where to next Define how Mary recognizes a reply type. Pull structured data out of incoming replies. Trigger Marketo updates and list adds automatically. Validate categorizations and extractions with the Test Suite. See every reply Mary has handled and why she classified it the way she did. Find specific emails across every category and field.