Power Platform Community Forum Thread Details

Hi Microsoft community,

We’re currently scaling an AI-powered solution with active clients and secured investor backing, and we’re seeking advice from those familiar with Microsoft's AI offerings. Specifically, we're evaluating whether Copilot Studio or Azure AI Studio would be the better platform for our next phase.

🔍 What We’re Building:
We’ve developed a suite of over 100+ AI assistants that adapt to each client’s needs using 5 core documents (like business overviews, brand guidelines, and SEO strategies). The goal is to provide ChatGPT-style interactions where users can seamlessly interact with AI agents that dynamically pull relevant data from these core documents to automate workflows across various departments (marketing, HR, finance, etc.).

🛠 Current Use Cases:
Here’s how our interconnected AI assistants collaborate to streamline business operations:

Researcher + Sales Guru + Sales Assistant + Executive Assistant:
- Conducts research, consults the Sales Guru to create a strategy, hands it off to the Sales Assistant to craft sales collateral, and uses the Executive Assistant to coordinate internal communications.
Report Creator/Data Analyst + Business Guru + Marketing Guru + Marketing Planner + Content Creator:
- Reviews engagement surveys, extracts insights, develops marketing strategies, creates detailed plans, and produces targeted content.
Marketing KPI Reviewer + Advisor + Planner + Content Creator:
- Analyses KPIs, provides strategic insights, builds marketing plans, and generates content to address identified issues.

💡 Our Challenge:
We aim to deploy these AI assistants to multiple customers simultaneously, each empowered by their own set of core documents. However, our current approach of manually importing configurations into each customer’s tenancy is not scalable, time-consuming, and exposes our IP to potential risks.

We’re looking for a solution that:

Enables scalable deployment of these AI assistants across multiple clients without manual imports.
Protects our intellectual property while providing seamless access to clients.
Supports ChatGPT-style interactions where AI agents dynamically utilise client-specific data from their core documents.
Allows for efficient integration of interconnected AI agents to automate workflows across different departments.

🔧 Key Questions:

Would Copilot Studio or Azure AI Studio be better suited to handle this type of scalable, secure deployment?
Which platform offers better support for document-driven AI interactions in real time?
How do these platforms differ in terms of scalability, customisation, and data security?
What other strategies or platforms could help us deploy AI assistants to multiple customers without the need for manual imports into each tenancy, thus mitigating the risk of exposing our IP?

We’d greatly appreciate insights from anyone familiar with Microsoft’s tools or any other strategies we could consider to address these challenges.

Thanks in advance for your help!

Categories:

General topics

Hello,

great question! I think that for your case, as the deployment is very big, I'd suggest that you should use Azure AI Studio instead of Copilot Studio. Why?

1. Copilot Studio is a low-code platform designed to create chatbots fast, thus hiding the complexity of deployment like: model which's used, model parameters etc.

2. Azure AI Studio is a tool for more complex business solution that need more customization - you can create model deployments by yourself, implement vector store/search from within Studio. More and more features are coming in so I'm not able to tell you everything. :-)

3. MOST IMPORTANTLY - your requirement to configure the environment automatically could be only fulfilled on Azure, there's even SDK already for the purpose of creating/managing the resources programatically:

https://learn.microsoft.com/en-us/azure/ai-studio/how-to/develop/sdk-overview?tabs=sync&pivots=programming-language-python

Overall Azure offers a lot of options to automatically create resources - PowerShell, az CLI, ARM templates...

4. In Azure itself you can implement robust security governance due to built-in RBAC, management groups, service principals and so on.

I'd definitely go with Azure AI Studio. Copilot Studio doesn't offer that kind of customization.