Build smarter and intelligent collaboration using Power Platform, Bots and Azure AI : The Concept

With many smart systems emerging in the market, users and organisations are now increasingly interested in using AI and other technologies to increase team productivity and providing efficient ways of getting work done. To this end, building smart collaboration between team members is imperative so users can focus more on productive interaction instead of redundant conversations.

Some of the key benefits of building smart collaboration are:

• Teams work together effectively while learning better ways to improve  
• Decrease redundancy by implementing smarter collaboration techinques  
• Decrease administration overhead by offloading work to systems 
• Consider integrations as your starting point with smart Applications

Solution Model and Flow

For achieving the above objectives, there are various approaches but simply put below is the high level approach to the solution model.

To implement the AI Integrations into user interactions, there are various ways we can interject the solution such as,

• Bots Conversations: Conversation AI for giving more of a human touch such as a Virtual assitant
• Smart AI Apps: Apps that use AI models to decipher information such as Form Processor
• Automation Proceses: AI Cognitive Services which automate processes without requiring data input

Any of the above solution we chose to implement as part of Smart Collaboration could be on its own or amalgation of all. Hence to simply the implementation process and make it easier for organisation to absorb the change, it is good to think about roll out in five stages as below.

The MVP (Minimal Viable Product) phase includes trialing most of the solution approaches to grasp the investment benefits of implementing the smart systems.

Solution Approaches

To enable the smart collaboration journey from a platform hosting point of view, below are four solution platforms we could look at.

1. Power Platform: As a start of the journey by using a Low code hosting platform. We will look at some of the offerings such as AI Builder, Power Virtual Agents in the upcoming blogs.
2. Bots: These are Conversational bots which provide a medium of human like interaction such as Bot framework solutions, Bot Composer and similar. We will look at more about business specific cases in more detail in upcoming blogs.
3. Extensions: With Automation solutions such as AI builder and UI flow, we can extend the automations to integrated systems
4. Virtual Assitants: This is an integrated Bot hosting that will allow multiple Bot implementations to talk to each other seamlessly without the user worrying about which Bot to use. We will also look at a more pratical implementation of this in upcoming blogs.

Each of the above areas will need specific discussions and hence we will look at each of them in separate blogs focussing on each area.

AI Design Principles

But before we move ahead, it good to understand one key aspect of AI systems, the process of designing any AI based application. Now, the some of the steps in the process might be different if we are working on a Bot vs a SaaS app but basically eventually the principles are same for each AI component within the solution.

Model

The key component of the AI design is the Model. The Model is the heart of the AI solution. The Model will determine the smart processing requirement we are seeking our solution to achieve, which can vary based on the medium such as Language vs Face vs Speech etc.

Train

After the model is ready, we train the model with real life like information so that it can detect entitties in real life implementation. During training we identify the attributes that we want our AI model to automatically detect. Normally it is advised that we train the model using lots of data and varied sets including real life examples so that it can extract information to near exact values.

Test

The next stage in the process is to test the model and make sure it works as expected with some real life examples. During the test process, we could also identify mismatches and additonal training items for making the model work with real life scenarios.

Implement

This is the final stage in the process and involves selecting one of the solution approaches (listed above) to implement the AI solution. This will also allow to collect any additional information that is missing in the model and later train/test the model.

By following the above AI design process, we will be able to effectively create a model for AI implementation

Conclusion

In this blog, we have seen the various aspects of building smart collaboration including a five stage approach. We have also looked at the high level principles of AI Design and solution building options with the various platforms.

In the upcoming blogs, we will look at some practical implementation of the AI solution using these platforms. We will also look at the high level approach for each of these platform implementations and quick tips when building the solutions.