Problem
Requests arrive through contact forms, inboxes, and internal channels. Without structure, someone has to read each message manually, judge urgency, send it to the right person, and remember what has already been handled.
What was built
The system combines web form and email intake into one normalised JSON shape. The AI layer returns category, priority, assigned team, summary, and confidence level. The ticket is stored in MySQL, and Slack receives a notification with enough context for the team to act quickly.
How the system works
- Webhook and IMAP intake land in the same data structure.
- AI returns strict JSON for category, priority, team assignment, and confidence.
- A MySQL record creates a stable ticket and a basis for history.
- Slack routing sends the context to the team, while edge cases stay on human review.
Operational value
The value is a steadier operating flow: every request gets the same shape, status no longer disappears in the inbox, and the team sees faster what needs attention and when a human decision is still required. No invented metrics are used because the original project material does not contain production client numbers.
HelpBot Ticket Desk shows how request intake can move from manual sorting into a controlled system. The key value is not the idea that AI decides everything. It is that uncertain cases go to review before the system makes the wrong routing decision.
