← All case studies

PrototypeIT Desk · voice and chat

Complete IT tickets first time, in the employee's own language.

Multinational services company · 4,000 employees · Prototype with live demo · not a named client

3languages: English, Hindi, ArabicReported

7guided questions, every timeReported

−60%follow-up questions per ticketTargeted

The problem

Employees of a 4,000-person multinational raised IT tickets through English-only forms.

£What it was costing them

Hours of back-and-forth per ticket, and slower fixes for everyone.

Tickets arrived with half the details missing.

The existing workflow

How the work was done, and where it broke.

Tickets arrived with half the details missing.

IT spent its first reply asking what was wrong.

Employees in other countries struggled with English-only forms.

The solution

Confirm before raising. The employee approves the summary before a ticket is created.

1Speak naturally

Employees describe the problem by voice or chat, in their own language.

2Ask what's missing

The assistant asks the seven questions IT needs.

3Read it back

It reads back a summary and raises the ticket once the employee agrees.

4Sync and report

Tickets sync to the service desk; managers see everything in an admin console.

Architecture

From input to outcome.

Inputs

Employee voice or chat, in English, Hindi or Arabic

AI layer

Asks the 7 questions IT needs
Reads back a summary

Human check

The employee confirms before a ticket is raised

Actions

Ticket raised and synced to the service desk

Outcome

Admin console for managers; email when resolved

Integrations

Works with the systems already in place.

⇄

IT service desk

⇄

Email notifications

⇄

Admin console

Business value

What changed for the business.

Complete ticketsReported

Every ticket arrives with the details IT needs.

Three languagesReported

English, Hindi and Arabic.

Fewer follow-upsTargeted

Designed to cut follow-up questions by 60%.

The evidence

Shown on real work.

An employee says in Hindi that their laptop won't connect to the VPN. The assistant asks the missing questions, reads back the summary and raises a complete ticket.

About the evidence

Prototype with a live demo. Languages and question flow are product features; the 60% figure is a design target.

How we label results

Have a process like this?

Tell us about it. The first discovery call is free.