Email Triage for Small UK Firms: A Classify, Draft and Approve Pattern

说明与用途

英国小型企业使用单一共享邮箱，邮件内容混杂。自动化模式对邮件进行分类、分配、草稿生成，并在发送前需人工批准，从而降低人工分类工作量并提升控制力。

问题

Small UK firms use a single shared inbox for diverse emails, causing owners to manually sort them each night.

结果

The pattern reduces manual triage, ensures human approval before sending, and provides data to refine snippets and classification.

工作流

• Classify email subject and body into one of seven categories using an LLM with a confidence threshold; uncertain cases default to "other".
• Route the email based on its category: assign to a team member, set a label, decide whether to draft a reply, and trigger alerts for complaints.
• Generate a draft reply for categories that allow drafting (e.g., new enquiries) using approved snippets; the draft is inserted into the thread but not sent.
• A human reviews the draft, edits if necessary, and sends the email; the decision is logged with message ID, category, confidence, approver, and edit status.
• Log every decision to identify heavily edited drafts and improve snippets.
• Test the classifier with anonymised real emails, including forwarded chains, replies without subjects, and attachment‑only messages, before trusting the system.

证据

Many small UK firms still run on one shared inbox: info@ or hello@. New enquiries, supplier invoices, complaints, spam and job applications all mix together, and the owner sorts them at night.

来源

The pattern is provider-neutral.

来源

We build this kind of triage after a 72‑hour audit of the inbox workflow.

来源