An AI assistant that reads, sorts, and prioritizes IT support tickets automatically — and shows the results on a live dashboard.
When something breaks at work — the WiFi drops, an app crashes, an order system fails — employees raise a "ticket" with the IT support team. A large company can receive hundreds of these every month.
Today, a person has to manually read each ticket, figure out what type of problem it is, decide how urgent it is, and send it to the right team. This takes roughly 6 minutes per ticket — about 50 hours of work every month for 500 tickets. Worse, when a ticket is accidentally sent to the wrong team, the fix gets delayed by hours.
This project hands that repetitive work to an AI assistant, freeing the human team to actually solve problems instead of sorting them.
For every support ticket that comes in, the AI automatically:
- Identifies the type of problem (network, hardware, software, account access, data integration, email)
- Decides how urgent it is — from "critical, fix now" to "minor, can wait" — using business rules (for example, a failure that stops customer orders is automatically treated as critical)
- Sends it to the right team
- Detects how the person is feeling (calm, frustrated, urgent) as an early warning sign
- Writes a one-line plain-English summary so the person fixing it understands the issue at a glance
All of this happens in about 3 seconds per ticket, instead of 6 minutes.
Tested on a realistic set of 500 sample tickets:
| What we measured | Result |
|---|---|
| How often the AI agreed with an expert human's judgment | 89% |
| Time to sort one ticket | ~3 seconds (down from ~6 minutes) |
| Urgent "critical" issues caught automatically | 44 |
| Average time issues took to resolve | 32 hours |
Why 89% and not 100%? Honesty matters more than a perfect-looking number. When the AI and the human disagreed, it was usually a genuine judgment call — and in several cases the AI was actually right, catching urgent order-system failures that a human had mistakenly marked as low priority. A tool you can trust is one that's measured honestly.
Instead of leaving the results in a spreadsheet, the project presents them on a clean, live dashboard that updates automatically. It shows:
- Headline numbers at a glance — total tickets, average resolution time, accuracy, and urgent-issue count
- An "Executive Analysis" written by the AI — a short plain-English summary of what the data means, with a recommended action (for example: "Email issues are the single biggest driver of volume — focus improvement efforts here")
- A breakdown of the most common problem types, so leaders can see where to focus
- A list of the urgent issues needing immediate attention
The standout feature is that the dashboard doesn't just show numbers — it explains them in words, the way a business analyst would. That turns raw data into decisions.
This isn't just a technical demo. It shows the full journey of a real business-improvement project:
- Identifying a costly, repetitive process (manual ticket sorting)
- Designing an automated solution that's faster and measurably reliable
- Building in honesty and safety — the system measures its own accuracy and sends anything it's unsure about to a human for review, rather than guessing
- Presenting results for decision-makers, not just technical staff
It combines process-improvement thinking (where to save time and money) with modern AI — exactly the kind of work that drives real transformation in a business.
To keep this safe to share publicly, the project uses made-up sample data, not any real company's information. It's built on everyday tools (a workflow automation platform, an AI service, and a spreadsheet) so the focus stays on the idea and the results. In a real company, the same approach would plug directly into the organization's existing IT ticketing system.
- The live dashboard
- The automated workflow (the "recipe" the AI assistant follows)
- The sample ticket data used for testing
- Screenshots of the dashboard
(for the technically curious)
Built with n8n (a visual automation platform), the Claude AI service for the intelligent classification, Google Sheets as the data store, and a custom web dashboard using Chart.js. Designed with Lean Six Sigma process-improvement principles throughout.
Built by Maria Corazon "Acey" Magallanes — Business Transformation Manager specializing in AI & automation. Lean Six Sigma Black Belt with ~15 years of experience across business process outsourcing, banking, telecommunications, and retail.