Case study · Internal IT

Routine IT requests handled in Teams, in minutes

IT helpdesks spend much of their time on routine requests: access, software, how-to questions. This is how we built an AI agent in Microsoft Teams that answers questions and completes approved actions through existing systems, with every step logged.

Start a project like this See more of our work

Project overview

An AI agent with tightly scoped actions.

Industry: Internal IT
Client: Confidential
Solution: AI helpdesk agent
Platform: Azure OpenAI + Teams
Timeline: 8–14 weeks
Team: 3 people
Actions: Pre-approved, scoped tools
Integration: ITSM, identity, software catalogue
Audit: Every action logged

Expected business outcomes

What the client set out to achieve

Measured by tickets resolved without an engineer and user satisfaction.

25–45% Of routine tickets resolved by the agent
Questions answered and approved actions completed automatically.

Target range; depends on request mix
Minutes To resolve routine requests
Instead of waiting in a queue for hours or days.

For requests within the agent's scope
100% Actions logged and scoped
The agent can only use approved tools with approvals where required.

Enforced by design

Business challenges

Where the client started

Engineers spend time on requests that follow the same steps every time.

Tickets

A large share are how-to questions or standard requests.

Waiting

Employees wait hours for simple access changes.

Knowledge

Answers in a knowledge base few people read.

Engineers

Skilled staff doing repetitive work.

Concern

Security teams worry about AI taking actions.

Proposed solution

What the agent covers

Answers from IT knowledge, and a small set of safe, approved actions.

Teams experience

Employees ask in Teams in plain language.
Knowledge answers

Answers from IT documentation with sources.
Scoped actions

Approved tools only: request software, reset MFA, request access.
Approvals

Manager or IT approval for sensitive actions.
ITSM integration

Tickets created and closed in the service desk tool.
Audit and limits

Every action logged; the agent cannot act outside its tools.

Project requirements

What the client asked us for

The request was to take routine work off engineers. Security shapes the design: the agent gets a short list of tools, each with limits and approvals.

Anything outside scope becomes a normal ticket with the conversation attached.

What the project had to deliver

01Answers in Teams
02Safe, scoped actions
03Approvals where needed
04Full audit trail

Key challenges & considerations

What made it harder, and what we did

Security review of every action

What we did

Each action was reviewed and approved by security before it was released.
Legacy systems without APIs

What we did

Systems without an API were left as handovers to an engineer, not automated unsafely.
Knowledge base quality

What we did

Knowledge articles were fixed where the evaluation showed wrong answers.
Change management for employees

What we did

The agent was introduced to one group at a time, with a clear route to a person.

What we watched for

Give the agent few, well-defined tools
Review every action with security
Hand over clearly when out of scope
Measure resolution, not conversations

Development process

Answers first, actions second

Actions are added one at a time after security review.

Stage 1
Request analysis

Ticket history analysed to find routine requests.
Stage 2
Knowledge answers

Agent answers questions in Teams.
Stage 3
First actions

One or two low-risk actions with approvals.
Stage 4
Security review

Each new action reviewed before release.
Stage 5
Expand

More actions added based on volume and risk.

Technology stack

Agents on Microsoft's stack

Models, tools and identity within the organisation's tenant.

AI

Azure OpenAI
Agent orchestration
Tool calling

Interface

Microsoft Teams bot

Integration

ITSM API
Microsoft Graph
Identity management

Governance

Action allow-list
Approvals
Audit logs

Quality assurance & deployment

Measured against a test set before launch

AI quality was measured, not judged from a demo. A test set of real examples with agreed answers decided whether the system was ready, and the same set was re-run after every change.

A test set of real examples with agreed answers
Accuracy thresholds agreed with the business before launch
Safety checks: refusals, permissions and prompt-injection tests
A person reviewing low-confidence output
Cost and response time measured at real volumes

Deployment

01Security reviewEach new action reviewed before release.
02ExpandMore actions added based on volume and risk.
03Pilot, then staged rolloutA small group of real users first, with results reviewed weekly before widening access.
04Monitoring and warrantyAccuracy, drift, usage and cost tracked after launch, and six months of free bug fixing.

Final deliverables

IT that answers straight away

Employees get answers and routine requests completed in Teams within minutes, and engineers focus on work that needs them.

Security sees every action the agent takes and every approval behind it.

Instant answers
Scoped actions
Approvals built in
Complete audit trail

Talk about a project like this

Handed over to the client

01Production AI serviceDeployed in the client's own cloud tenant
02Source code and modelsIn a repository the client can access
03Test set and resultsThe evidence behind the launch decision
04MonitoringAccuracy, usage and cost
05Six-month warrantyFree bug fixing after delivery

Success metrics

How success is measured

Measure	Outcome	How it is measured
Of routine tickets resolved by the agent	25–45%	Target range; depends on request mix
To resolve routine requests	Minutes	For requests within the agent's scope
Actions logged and scoped	100%	Enforced by design

Verified reviews

What clients say about our custom software work

Verified reviews from our clients on the same kind of work, published on Clutch.

5.0 27 verified reviews

5.0 Custom Software Development

Web-Based Coupon App Development

The work produced by the team was of exceptional quality.

Jon Thies CTO, Model Rocket Feb 2022 – Mar 2023
Verified review on Clutch (Jon Thies, opens in a new tab)
5.0 Custom Software Development

Mobile & Web Platform Development & Design

Silver Scintilla was responsive, respectful, and enjoyable to work with, quickly addressing any questions or requests.

DeShawn Brown CEO & Founder, Lithios Jun 2021 – Nov 2022
Verified review on Clutch (DeShawn Brown, opens in a new tab)
5.0 Custom Software Development

Web-Based HRM System Development

Silver Scintilla provided the best solution with scalable functionality, personalized features, and robust security.

Suheb Khan Director, Collaborative Insight Technologies Jul 2020 – Jun 2022
Verified review on Clutch (Suheb Khan, opens in a new tab)

See All Testimonials Read All Reviews on Clutch

Routine IT requests handled in Teams, in minutes

What the client set out to achieve

Where the client started

What the agent covers

Teams experience

Knowledge answers

Scoped actions

Approvals

ITSM integration

Audit and limits

What the client asked us for

What made it harder, and what we did

Security review of every action

Legacy systems without APIs

Knowledge base quality

Change management for employees

Answers first, actions second

Request analysis

Knowledge answers

First actions

Security review

Expand

Agents on Microsoft's stack

AI

Interface

Integration

Governance

Measured against a test set before launch

IT that answers straight away

How success is measured

What clients say about our custom software work

Services we used on this project

AI & ML development

AI proof of concept

AI agent development

Have an idea? Let's turn it into AI-powered software.