ArchitectAI for SharePoint

#sharepoint #ai

Imagine a new addition to your SharePoint toolbox: ArchitectAI for SharePoint. This innovative application leverages the power of Generative AI (Gen AI) to go beyond just answering questions.

ArchitectAI for SharePoint acts as your conversational guide, helping you design an optimized information layout for your SharePoint Online site.

The beauty of ArchitectAI for SharePoint? It even generates a bit of PowerShell code to streamline the setup process for administrators and advanced users.

I hope the prospect of building a conversational assistant like ArchitectAI for SharePoint, or a similar application using the techniques outlined here, excites you as much as it did me while developing it!

ArchitectAI for SharePoint - YouTube

ArchitectAI for SharePoint: Prototype Showcase - AI-Powered Information ArchitectureTake a sneak peek at ArchitectAI, a revolutionary conversational app desi...

youtube.com

Prompt Engineering

Building effective Gen AI applications hinges on providing the right background information. Think of it as giving Gen AI a clear roadmap. We need to tell it:

Its Mission: What's the task at hand? Is it generating content, making recommendations, or something else entirely?
The Information Highway: What specific details does Gen AI need to access to complete the task?
The Destination: What kind of answer are we expecting? Structured data (like JSON), plain text, or something else?

Fortunately, Gen AI is multilingual – it understands various prompt formats. One popular method is TAG (Task, Action, Goal):

Task: The overall objective (e.g., "Generate document outlines").
Action: How Gen AI should achieve the task (e.g., "Analyze keywords").
Goal: The desired outcome (e.g., "Create outlines with relevant sections").

While TAG is a powerful tool, there are other options like APE, CARE, TRACE, etc. (These are specific prompt formats and exploring them individually might be outside the scope of this blog post).

Specifying the Answer Format is Key

Crucially, we need to be very clear about the format we want the answer in. Remember, our application interacts with Gen AI, not us directly. For ArchitectAI for SharePoint, the desired answer format might be:

{
    explanation: string,
    architecture: {
        site: 'Communication site' | 'Team site',
        lists: [
            {
                name: string,
                description: string,
                columns: [
                    {
                        name: string,
                        description: string
                    }
                ]
            }
        ],
        libraries: [
            {
                name: string,
                description: string,
                columns: [
                    {
                        name: string,
                        description: string
                    }
                ]
            }
        ]
    },
    question: string,
    code: {
        powershell: string
    }
}

By providing clear instructions and specifying the desired answer format, we empower Gen AI to deliver the most valuable and actionable results for applications like ArchitectAI for SharePoint.

What else

We've discussed how crafting clear prompts is crucial for guiding Gen AI to generate the information we need. However, the power of Gen AI extends far beyond simply providing answers to questions.

Think beyond question-and-answer formats. Imagine an application where you can have a natural conversation with Gen AI to troubleshoot a technical problem. You describe the issue, and Gen AI asks clarifying questions, analyzes the information, and guides you through solutions in a step-by-step dialogue.

Learning and training can also benefit from the conversational approach. Imagine a platform that personalizes content and explanations based on your specific learning style. Through conversation, Gen AI assesses your knowledge level, identifies areas for improvement, and provides tailored resources, making the learning process more engaging and effective.

These are just a few examples, and the possibilities are constantly expanding. As Gen AI continues to evolve, we can expect even more innovative ways to leverage the power of conversation in building applications that are not only informative but also engaging and interactive.