What Makes Emely Unique

A quick foreword from our Founder

At the core, we built Emely to help businesses with process improvement.  Like many other software products, we've incorporated the latest technologies in our build, and will continue to do so, to make our solution as effective as possible for it's ultimate purpose; help businesses drive staff productivity up by saving time.

So - What’s unique about Emely?

• The platform is 100% Microsoft Azure
• There is zero effort required for end users to start using it - she automatically shows up in Microsoft Teams.
• Responses to natural language-based questions are based on information controlled with Azure RBAC security permissions
• Questions can be mapped to automations which are controlled and developed by the in-house IT department through Microsoft Power Automate.
• We give end users not familiar with accessing the GPT chat model the opportunity to ask Emely (through an embed chat within the Microsoft Teams chat) to have an open AI-based discussion, but only after vetting that the previous Q&A and knowledge base look ups were insufficient for the answer.

Only giving insights about my company’s data is inherent to any LLM.  What's different here?

We have applied OpenAI's API to not only look at large sets of data, but we leverage a hierarchy of information retrieval which allows for specific answers to override the assumptions of the AI.  This allows for specific delivery of "correct" information instead of any AI's best guess.  This is a very important detail for standardizing execution of routine business processes.

How does it fact check itself to tell me “truths”?

The AI does not necessarily check itself for facts, as it isn't necessarily aware of what is true and is not. However, the workflow starts by matching a question and answer pair first which then returns back a very specific answer.  The intention here is to make sure that we capture the users question and match it to an answer that a company can state as true so it always overrides the "guesswork" of the AI against the larger data set and it's algorithms answer.

Does it not do Cartesian mapping and measure distance like other GPT?

According to our understanding of the documentation, there is no Cartesian mapping and distance measurements which are a part of the algorithm

Is this just a "wrapper" service around the already-existing OpenAI or ChatGPT?

No - we don't ever touch the public OpenAI Chat API. We use an internal API to leverage a GPT-4 LLM hosted in the Azure OpenAI service to perform all our natural language processing. This is important because it emphasizes the security of our application architecture.

How is this any different from ChatGPT or Bing Chat?

Both ChatGPT and Bing Chat are built on the same platform; OpenAI, which Emely is also built directly on.  In that regard, all three are peers.  However, Emely AI is built out with a robust feature set specifically intended to meet the business purpose of delivering company proprietary information, automations and directly address the day-to-day needs of the modern information worker.

Special Note: Kevin (Lead Developer)'s Feedback

Something that makes Emely unique is that it isn't just an LLM; Emely is a multifaceted AI application that uses many different AI algorithms to accomplish a variety of tasks that would otherwise require specialized software to automate: 

• It can build and maintain a company knowledge base that can be accessed through a chat
• It can collect insights about its usage as well as produce reports that can be requested on-demand via chat
• It can maintain a database of custom-built automations that can be triggered by key words & phrases in a chat. 
• t can also infer the intent behind user's questions and suggest automations that could help

Would using Azure OpenAI service in conjunction with features of Azure Cognitive Services such as Cognitive Search and Cognitive Language Understanding be an example of using a Cartesian mapping framework to integrate KGs?

"Using Azure OpenAI Service in conjunction with Azure Cognitive Services such as Cognitive Search and Cognitive Language Understanding is not an example of using a Cartesian mapping framework to integrate Knowledge Graphs (KGs)." - Special note: This particular bit of information comes directly from Bing Chat.

In this setup, Azure OpenAI Service uses powerful language models like GPT-3.5-Turbo and GPT-4 to provide responses based on your data. Azure Cognitive Search is used to determine what data to retrieve based on user inputs and provided conversation history. The retrieved data is then augmented and resubmitted as a prompt to the OpenAI model. This is a form of mapping, but it’s not Cartesian in the traditional sense.

The integration of Azure OpenAI Service with Azure Cognitive Services enhances the capabilities of the AI system by allowing it to leverage both structured data (from KGs) and unstructured data (from text), but it does not involve Cartesian mapping.  So I think the short answer to his question is no, Emely doesn't use a Cartesian mapping framework for integrating Knowledge Graphs (KGs).