Microsoft has done a lot to extend the “copilot” concept of coding into new areas. And at the Build 2023 conference, Microsoft leadership announced new capabilities in Azure AI Studio that enable individual developers to create their own co-pilot. This news is interesting because it will allow engineers to create co-pilots who are more knowledgeable about certain areas.
Below, we’ll cover some of the key takeaways from the Microsoft Build keynote on Tuesday, May 23, 2023, and explore what the announcement means for developers. Explore the copilot stack to see why you might want to build your own copilot.
What is a co-pilot?
Copilot is an artificial intelligence tool that assists with cognitive tasks. So far, the Copilot idea has been mostly associated with his GitHub Copilot. GitHub Copilot debuted in late 2021 to bring real-time automated suggestions to code editors.
“GitHub Copilot is the first solution built using the new transformative large-scale language model developed by OpenAI, and Copilot works with all popular programming languages, dramatically accelerating productivity. We will provide an AI pair programmer to work with,” said Scott Guthrie, the company’s executive vice president. Microsoft.
However, Microsoft recently launched the Copilot X with the GPT-4 model. New features also provide chat capabilities using GitHub Copilot Chat, which accepts prompts in natural language.
But the Copilot craze doesn’t stop there. Microsoft is actively integrating Copilot into other areas such as Windows and Microsoft 365. This means end users can create natural language prompts to spin up documents across Microsoft suites such as Word, Teams, and PowerPoint. other applications. Microsoft also builds Dynamics 365 Copilot, Power Platform Copilot, Security Copilot, Nuance, and Bing. It is easy to imagine that co-pilots will be active in many other development environments as well, riding on this momentum.
Having built these copilots, Microsoft began to realize that they had something in common. This led to the creation of a common framework for building copilots built on Azure AI. In his Build, Microsoft revealed how developers can use this framework to build their own copilot.
Build your own co-pilot
The underlying AI models are powerful, but they can’t do everything. One limitation is often no access to real-time context or private data. One way around this is to extend the model through plugins with REST API endpoints to get context for the task at hand. With Azure, you can do this by building ChatGPT plugins within VS Code and GitHub Codespaces to connect your apps and data to AI. However, you can take this even further by creating your own co-pilot or leveraging his bespoke LLM.
Understand the Azure Copilot stack
Part of the Azure OpenAI service is the new Azure AI Studio. The service allows developers to combine AI models such as his ChatGPT and GPT-4 with their own data. This can be used to build a more intelligent and situational co-pilot experience. Users can leverage open source LLM, Azure OpenAI, or bring their own AI models. The next step is to create a “metaprompt” that provides a role for how the copilot works.
So what’s the process like? First, create a new Azure OpenAI resource. Next, add a data source. A data source can be as simple as a PDF file or Word document. You can then choose to have the Generative AI expand the response or be limited to the data you provide. Services are then exposed through APIs and injected into any application format. The result could be a chat assistant that loops on your private data and works well with your own apps and data.
Microsoft has also built tools to keep private data safe and encrypted throughout the AI orchestration process. “AI orchestration includes foundation building, rapid design and engineering, evaluation, and AI safety. said Guthrie.
you get a co-pilot you Get a co-pilot…
The co-pilot explosion comes as more companies integrate LLM and generative AI into their products. Among these powerful AI additions is the ability to accelerate low-code development with advanced autocomplete functionality. And the Azure OpenAI service appears to be a promising platform for quickly launching custom copilots.
“AI will profoundly change the way we work and the way all organizations operate, all existing apps will be reinvented with AI, and new apps that were not possible before will be built with AI.” he said. Guthrie.
By building their own co-pilot, organizations use generative AI to customize experiences, give more context to specific domains, and solve unique pain points. And integrating the co-pilot into the realm of the end-user interface helps make AI more accessible in a broader sense.
“Software has eaten the world. Now it’s AI’s turn,” said GitHub CEO Thomas Dohmke.
