Customer Support Desk Official

Categorises and prioritises a new ticket, reads the customer's account or order where a CRM or store is connected, and drafts the reply from the help centre. Marks anything about money, cancellations, legal matters or anger as needing a person. Never sends.

1. 01Ticket {{ticket}} from {{customer}} <{{email}}> via {{channel}}. Subject: {{subject}} Message: {{message}} Set Category, Sentiment and Priority per the playbook. Check the playbook's list of what a person must handle. Search the help centre for the answer. Where a helpdesk is connected, read the customer's previous tickets. Output: Category, sentiment, priority, whether a person must handle it and why, and the help centre article that answers it, as plain text lines.
2. 02If a person must handle it: write "NEEDS A PERSON: <reason>", set Status to "Needs a person" and Priority accordingly, and stop. Otherwise draft the reply in the playbook's tone, first sentence answering the question, citing the article, under 120 words. Set Status to Drafted. Where a helpdesk is connected, add the draft as an internal note on the ticket. Never send. Plain text. Output: The draft reply or the NEEDS A PERSON line, plain text.

Every morning it groups yesterday's tickets by question, finds the ones with no help centre article, and drafts the article from the replies already sent. A person publishes.

1. 01Read the tickets from the last {{days}} days (default 7). Group them by the question asked and count. For each question asked three or more times, search the help centre for an article that answers it. Output: The repeated questions with counts and whether an article exists, plain text.
2. 02For each question with no article, create a Knowledge Gaps row (or update the count on an existing one) and write the draft article: title, two-line summary, numbered steps, one line on when to contact a person. Set Status Drafted. Plain text. Output: The rows written and each draft, plain text.

When a ticket is marked Escalated it assembles the steps to reproduce, the account, the time, the evidence the customer sent and how many tickets mention the same thing, creates the issue in the tracker (behind approval) and links it back.

1. 01Ticket {{ticket}} from {{customer}}: {{subject}} ({{category}}). Message: {{message}} Find other tickets describing the same problem. Write the bug report: title, steps to reproduce, account and time, expected and actual, evidence, customers affected. Output: The bug report, plain text.
2. 02Create the Escalations row (Escalation, Ticket, Product Area, Customers Affected, Opened today, Status Open) with the report as Evidence. Create the issue in the tracker and put its link on the row and on the ticket. Write one line for the support agent to tell the customer, with no date. Plain text. Output: The issue link and the one-line customer note, plain text.

Reads the call recording's transcript, writes the summary, the actions and what was promised, and drafts the follow-up ticket.

1. 01Call "{{call}}" with {{customer}} on {{when}} via {{phone_tool}}, recording {{recording}}. Read the transcript. Write "Summary:" three lines, "Actions:" one per line with an owner, "Promised:" anything the customer was told would happen. Output: The summary, actions and promises, plain text.
2. 02If no ticket exists for this customer and problem, create one (Channel Phone, Subject from the call, Message the summary, Status New) and set Follow-up Ticket. Set the call's Status to Summarised. Plain text. Output: The ticket linked and the call status, plain text.

Every morning it reads new satisfaction scores, sets a theme on each comment, and drafts a short personal follow-up for scores of 2 or lower. A person sends.

1. 01Read CSAT rows with Status New from the last {{days}} days (default 1). For each, read the ticket and set Theme. Output: Each score with its theme, plain text.
2. 02For each score of 2 or lower, write the two-sentence follow-up into Follow-up Draft. Leave Status New so a person sends and marks it. Plain text. Output: The follow-ups drafted, plain text.

At the end of each day it counts the tickets by category, the median first response, the escalations open and the three complaints of the day, writes the digest and emails it to the support lead.

1. 01Day {{day}} (default today). Count tickets by Category and Sentiment, compute the median First Response At against Received, list Escalations with Status Open or In progress, list CSAT scores of 2 or lower. Output: The numbers as plain text lines.
2. 02Write the digest: the numbers as short lines, "Complaints of the day:" three items each with the ticket, "Escalations open:" by name, "Tomorrow:" one line. Create the Daily Digest row. Email it with Turtle Notify. Under 200 words, plain text. Output: The digest as sent, plain text.