Client engagement · IT services

AI agent for service-desk triage

An AI agent that reads incoming tickets, classifies and routes them, and drafts first responses for a technician to approve.

01

Overview

An IT service provider wanted to spend less technician time on sorting tickets and more on solving them. Every new ticket had to be read, categorised, prioritised and routed before real work could begin, and first responses varied in quality.

We designed and built an AI agent that triages incoming tickets and drafts a first response for a technician to approve, with governance, security and measurement built in from the start.

02

Starting point

Technicians spent the first part of every ticket working out what it was about and who should own it. Quality varied with who happened to be on shift.

Tickets arrived by email, portal and Teams, often with screenshots and incomplete information. Routing rules lived in people’s heads, so accuracy depended on who was on shift.

The organisation was open to AI but had clear conditions: no customer data used to train public models, a human decision before anything reaches a customer, and a full audit trail.

03

Objectives

  • Classify, prioritise and route tickets consistently from the moment they arrive
  • Draft accurate first responses grounded in the organisation’s own knowledge base
  • Keep a human in control of every customer-facing action
  • Protect personal and customer information end to end
  • Measure quality continuously and catch regressions before users do

04

Our approach

  1. 01

    Discovery

    Analysis of historical tickets and routing rules with senior technicians, definition of categories and priorities, and a governance checklist agreed with leadership.

  2. 02

    Design

    A narrow agent role: read tickets and the knowledge base, suggest category, priority and queue, and draft a reply. No ability to close tickets, change systems or email customers directly.

  3. 03

    Build

    Retrieval over approved knowledge articles, prompt and tool design, redaction of personal information before model calls, and integration with the ticketing workflow and Teams.

  4. 04

    Evaluate

    A test set built from real, anonymised tickets with known correct answers, run on every change to prompts, models or tools, with accuracy and escalation tracked over time.

  5. 05

    Roll out

    Shadow mode first, where suggestions were logged but not shown, then assisted mode for one team, then wider rollout as quality targets were met.

05

What we delivered

  • Ticket classification and routing by category, priority and skill
  • Drafted replies grounded in the knowledge base, sent only after human approval
  • Least-privilege service identity, full audit trail and redaction of personal information
  • Evaluation set of real tickets, re-run on every prompt or model change

06

Security, compliance and risk

Human in the loop

Every draft reply requires technician approval, and the agent’s suggestions are shown with the reasons behind them.

Security

The agent runs under its own least-privilege identity, cannot reach systems outside its role, and treats ticket content as untrusted input to limit prompt-injection risk.

Privacy and audit

Personal information is redacted before processing, model data is not used for training, and every suggestion, approval and change is logged.

07

Outcome

Tickets reach the right queue from the start, first responses are consistent, and every AI action is logged and reviewable.

08

What happens next

The evaluation suite runs on every change, knowledge articles are refreshed as gaps are found, and new workflows are added only once the current ones meet their quality targets.

More projects

All projects
Live
Our platform · Bhargava Group · 2026

Alliance corporate platform

Our own bilingual corporate platform: five practice sites, projects, insights, search and lead capture on an edge-hosted, near-zero-JavaScript architecture.

Software & AIDigital GrowthCloud
Managed service
Client engagement · Manufacturing

Managed IT and security for a multi-site manufacturer

One accountable IT team for several plants and a head office: monitoring, patching, identity security and a helpdesk that knows the shop floor.

Managed ITCloud
Delivered
Client engagement · Professional services

Microsoft 365 hardening and Copilot readiness

A tenant secured and cleaned up before AI arrived, so Copilot could be switched on without surfacing what it should not.

Managed ITSoftware & AI

Talk to us

Tell us where you want to be in twelve months.

A 30-minute consultation with a senior consultant. You leave with a clear next step — whether or not it involves us.

Privacy choices

Choose which optional technologies we may use. You can change this at any time from “Privacy choices” in the footer. Cookie policy

Strictly necessary

Security, spam protection (Cloudflare Turnstile) and remembering these choices. Always on.

Always on