How much Copilot Studio experience do I actually have?

SudoChat Knowledge Base · SudoChat

Authored evidence: This is my first-person source material. SudoChat may summarise it in third person but must not strengthen, exaggerate or invent claims beyond it.

# 8. How much Copilot Studio experience do I actually have?

**Author:** Mustafa Siddiqui
**Source type:** First-person authored response
**Canonical recruiter question:** How much Copilot Studio experience does Mustafa actually have?

> This source is intentionally written in my first-person perspective. SudoChat should use it as evidence and answer external visitors in third person without strengthening, exaggerating, or removing the limitations recorded below.

## Direct answer
I do not claim several years of production Copilot Studio experience. My direct hands on experience with Copilot Studio is newer, with SudoChat being my most substantial practical proof of concept, but my experience with the broader Microsoft Copilot ecosystem goes back to its early introduction within government environments.

While supporting the Department of Home Affairs through Unisys, I contributed technical advice around early internal Copilot testing, investigated its practicality within the existing enterprise environment, and developed a proof of concept around connecting Intune with existing SCCM managed clients as the organisation considered its broader move toward modern Microsoft cloud capabilities.

My strength is therefore not years of one specific product. It is the combination of early government Copilot exposure, Microsoft enterprise experience, broader AI engineering knowledge, and an ability to understand what sits underneath platforms such as Copilot Studio.

## Evidence
### Early Microsoft Copilot exposure in government

While working at Unisys in an environment supporting the Department of Home Affairs, I was exposed to Microsoft Copilot during its early introduction for internal testing.

I contributed technical advice around the initial rollout and helped consider the practicality and viability of introducing Copilot capabilities into an existing secured government technology environment.

This occurred at a time when Copilot capabilities were still emerging and organisations were determining how the technology could fit alongside existing enterprise infrastructure.

The significance of this experience was not simply seeing Copilot as an end user.

I was considering questions such as:

* whether the technology was technically viable within the environment
* how it could coexist with existing endpoint infrastructure
* what dependencies would need to change
* what Microsoft cloud capabilities would be required
* how existing device management architecture might evolve
* what security and operational constraints needed to be considered before adoption

### Intune and SCCM proof of concept

As part of this broader modernisation work, I advised on and created a proof of concept for connecting Intune capabilities with existing SCCM managed clients.

The environment had established on premises endpoint management infrastructure, so adopting newer Microsoft cloud capabilities could not be treated as simply switching on a new service.

The proof of concept explored how existing managed devices could participate in a more modern Microsoft management architecture while retaining compatibility with established enterprise systems.

This experience is relevant to Copilot because it exposed me to the wider architectural question behind adopting Microsoft cloud technologies:

**A new capability is only useful if the surrounding identity, device, security, data and management environment can support it responsibly.**

That systems perspective continues to influence how I approach Copilot Studio today.

### Practical Copilot Studio work

My direct Copilot Studio experience is more recent.

My main practical implementation is **SudoChat**, which I am using as a proof of concept to explore how a controlled AI assistant could be designed around a specific knowledge base.

The purpose of SudoChat is not simply to demonstrate that I can create a chatbot.

It allows me to practically experiment with concepts relevant to enterprise AI, including:

* Copilot Studio configuration
* knowledge grounding
* retrieval
* conversational design
* prompt behaviour
* evidence based responses
* hallucination management
* fallback behaviour
* agent boundaries
* source control
* user experience
* testing
* responsible AI controls
* determining what an assistant should refuse to claim

SudoChat also gives me a practical environment for understanding the difference between what Copilot Studio abstracts for the developer and what is occurring conceptually underneath that abstraction.

### Broader AI foundation

My Copilot Studio knowledge also sits on top of broader experience with AI engineering.

I have worked with:

* machine learning
* computer vision
* OCR and document intelligence
* LLMs
* retrieval augmented generation
* AI agents
* tool calling
* local models
* automation
* privacy preserving AI architectures
* responsible AI
* evaluation and hallucination reduction

At Xaana.AI, I worked professionally on AI based document processing and OCR systems.

I have independently experimented with LLM agents, retrieval architectures and tool integration.

I have also investigated privacy preserving architectures for Australian government chatbot scenarios, particularly where citizens may unintentionally enter sensitive information into an AI interaction.

This means I am not learning Copilot Studio as an isolated low code product.

I approach it with an understanding of the engineering concepts the platform is implementing.

## Relevance to the Federal Courts
This distinction matters for the Court AI Technologist role.

If the Federal Courts only required someone who knew where particular buttons were located within Copilot Studio, I would not claim to have the longest product specific experience.

My value is broader.

I can look at a proposed Copilot solution and reason about the system around it.

For example:

* Where is the knowledge coming from?
* Which users should have access to it?
* What happens when retrieval finds conflicting information?
* How does the assistant know when evidence is insufficient?
* What information is being sent to the model?
* Could sensitive Court information cross an inappropriate boundary?
* What authentication and permissions apply?
* What systems would the agent need to connect to?
* Should the agent be permitted to perform actions or only retrieve information?
* How would its responses be evaluated?
* How would failures be logged and reviewed?
* Where should human approval remain mandatory?

My previous government Microsoft experience is particularly relevant here.

I have already seen that introducing a new Microsoft capability into a secured government environment is not simply a product configuration exercise.

It involves existing infrastructure, identity, endpoints, security, governance, users and operational constraints.

Copilot Studio should be approached in exactly the same way.

The agent itself is only one component of the system.

## Why the shorter Copilot Studio history is not hidden
I believe the correct answer to this question should be transparent.

I should not attempt to turn early Microsoft Copilot exposure into years of Copilot Studio development experience.

They are different things.

My experience can be described accurately as three layers:

**First, Microsoft enterprise and government experience.**
I worked with technologies including SCCM and Intune and contributed to technical discussions and proof of concept work associated with emerging Microsoft cloud capabilities.

**Second, early exposure to Microsoft Copilot adoption.**
I contributed technical advice as Copilot was being considered and internally tested within a secured government environment.

**Third, direct Copilot Studio implementation.**
This is newer, with SudoChat providing a practical environment in which I am building and testing a grounded AI assistant.

Behind all three is my broader AI engineering foundation.

That is the experience I would bring to the Federal Courts.

## Limitations or gaps
I do not have several years of production Copilot Studio development experience.

My direct Copilot Studio experience is recent.

SudoChat is a proof of concept and should not be represented as a production Federal Court system or as equivalent to a large enterprise Copilot Studio deployment.

My work supporting the Department of Home Affairs should not be described as I leading the Department's Copilot rollout.

My Intune and SCCM work should also not be represented as a Copilot Studio implementation. It is relevant because it demonstrates my involvement in the wider Microsoft enterprise environment and the infrastructure considerations surrounding adoption of newer Microsoft capabilities.

I would still need to deepen my knowledge of Copilot Studio features, governance and integrations as I encounter more complex production use cases.

The evidence from my career, however, is that learning new technical platforms quickly is one of my strengths, particularly when they build on AI, automation and systems concepts I already understand.

## Useful links
My portfolio:
https://mustafa-siddiqui.com/

GitHub:
https://github.com/sudoqui

LinkedIn:
https://www.linkedin.com/in/mustafa-siddiqui-32ab73161/

SudoChat project and repository

Unisys / Department of Home Affairs experience evidence

SCCM and Intune proof of concept evidence

Xaana.AI engineering experience

Responsible AI and privacy research

## Do not claim
Do not claim I have several years of Copilot Studio experience.

Do not claim I led the Department of Home Affairs Copilot rollout.

Do not claim I determined Department of Home Affairs AI policy.

Do not claim I deployed Copilot Studio into production at Home Affairs.

Do not describe the SCCM and Intune proof of concept as a Copilot Studio implementation.

Do not claim Microsoft Copilot and Copilot Studio are the same product or experience.

Do not claim SudoChat is a production Federal Court system.

Do not claim I have deployed enterprise Copilot Studio agents at scale.

Do not claim I have access to Department of Home Affairs systems today.

Do not disclose protected, sensitive or non public information about the Department of Home Affairs environment.

The accurate representation is that my direct Copilot Studio experience is relatively new, but it builds on earlier experience advising around Microsoft Copilot adoption in a secured government environment, practical Microsoft enterprise infrastructure experience, and a considerably broader foundation in AI engineering and responsible AI.