Let’s dive into a real custom copilot application example from some of our recent development. This is part 2 of our exploration into custom copilot applications. In part 1, we looked at the burgeoning AI landscape and where custom applications can fit in to increase productivity and make organisations operations more effective.
One such challenge was presented to us by a client looking to stay ahead in a competitive environment. They were part of an industry where winning bids could pivot on having the most insightful analysis and presentation of data. Their bidding team needed an edge—a way to sift through extensive archives of project data to craft compelling and informed bids. The client was aware of AI’s potential but needed a partner to explore and identify practical AI-driven solutions to reinforce their bidding process.
Transparity stepped in with a focus on creating a custom AI solution that could empower their bidding team. The journey we embarked on with our client involved navigating the possibilities of AI to find and tailor applications that could turn their vast data into winning strategies without compromising the agility and security of their operations.
Let’s delve into this recent journey of creating a custom copilot application example for a client and see what we can learn.
What the client wanted
Our journey began with a client who like many these days wanted to see how they could take advantage of the burgeoning AI revolution to empower their teams to get more done with less effort. They were navigating uncharted territory—the intersection of evolving technology and uncertain user reception and utilisation patterns.
To tackle this challenge, they knew they wanted an iterative approach where they could take small steps one after the other, adjust tactics or direction with each step to make sure they didn’t step too far only to find they’d missed the mark because the AI landscape had changed or they didn’t fully understand how their teams could use the new tools.
This was a great choice and one we would highly recommend. If you are going to fail, fail early and fast. To that end, we started by developing a very basic custom copilot application. In short order, they had the application in their hands and providing some benefits, even if it wasn’t yet ticking all the boxes. Still, they were getting feedback on what was working and what wasn’t.
We took that feedback and worked with them to make adjustments and to take advantage of updates to the underlying Azure AI Services. This provided new benefits with little effort as at the time we started the project, it would have taken considerable effort or not been possible at all.
The solution: Basic AI Chat Copilot
Our client wanted a custom AI Chat application. They couldn’t use the available public offering at the time due to security and compliance concerns and they also knew that down the line they would want custom features. The basic AI Chat copilot we developed for them was a significant leap forward at its launch, featuring the ability to manage multiple concurrent conversations—a stark contrast to the linear chat interfaces of the time. Users could effortlessly pause and resume dialogues, ensuring no loss of context or continuity when tackling complex tasks.
In addition to conversation management, the copilot was equipped with a metadata feature, allowing users to enrich each interaction with additional information such as project numbers and tags. This function proved critical for organising and attributing conversations, facilitating quick retrieval and better data management in future searches. It became a powerful tool for the client’s bidding team, who could now navigate and utilise their extensive project data with unprecedented ease and efficiency.
As we continue to refine the AI Chat copilot, these advanced features will be honed further, and new capabilities will be integrated. Our commitment is to evolve the application in sync with our client’s needs and the rapid progression of AI technology, enhancing the user experience with every update.
The Solution: Document Search Copilot
Our client also wanted a way for their users all over the world to be able to find company documents relating to a topic of interest even if those documents were on the other side of the globe and the contents were not accessible to them for security reasons. We developed a Document Search Copilot application for navigating a secure, globally distributed document repository. To comply with stringent security protocols and ensure document confidentiality, the system was engineered to disclose the existence of documents matching user queries without revealing their content.
Leveraging advanced natural language processing, the Document Search Copilot allows users to conduct searches through conversational queries, responding with relevancy rankings for existing documents. This approach ensures that sensitive information is never compromised during the search process. Users are informed about the presence of relevant documents, which they can access through established company procedures, ensuring all interactions adhere to security and compliance regulations.
By prioritising security in its design, the Document Search Copilot provides users with the necessary tools to identify pertinent documents efficiently, without direct exposure to the content. This careful balance between accessibility and protection is central to the system’s effectiveness, especially in environments where information sensitivity is paramount.
Future enhancements to the copilot will continue to refine its search capabilities while upholding the core principle of secure document discovery. The focus will remain on delivering an intelligent, user-friendly search experience that aligns with the rigorous security demands of our clients’ operational landscapes.
We will also leverage this document search capability as part of more complex AI copilot workflow solutions which assist in more specialised tasks.
The Next Phase of Our Journey
I hope this look at a real custom copilot application example helps you get a feel for what is possible with custom AI. Our journey with this client is far from over. As we venture into the next phase of development, our focus is on harnessing the potential of the Azure AI Assistants API, delving into the capabilities of Kernel Memory, and integrating the advanced features of the latest AI models, such as the groundbreaking GPT-4o. These cutting-edge technologies represent the frontier of AI development, and we are poised to explore their full range.
The fusion of these technological advancements will enable our copilots to perform with a new level of sophistication and customisation. With these tools, we aim to unlock creative potentials and workflow efficiencies that could redefine productivity in numerous industries. The future we’re building towards is one where our AI solutions are not just tools, but partners in innovation.
