Since their inception in 2020, Generative AI and AI-assisted tools have taken the world by storm, doubling in market size year over year. According to management consulting firm Oliver Wyman, more than 55% of global workers are already using generative AI at least once a week. Further integrations into the daily workflow of users are forecast to skyrocket productivity gains, generating savings of upwards of 300 billion hours collectively by 2030.
Exciting new advancements in Copilot unveiled during the Microsoft Build conference in May have only fueled expectations, with updates focused on custom developer-crafted Copilots, enhanced extensibility features, and personalized access to Copilot from a single interface on Microsoft Teams. These innovations are expected to further streamline workflows, foster increased collaboration, and drive productivity across all sectors.
To gain a better understanding of what these groundbreaking announcements mean to the way we work, let's delve deeper into how Microsoft Copilot continues to evolve as an indispensable tool for modern enterprises.
Developers rejoice! Introducing AI Studio
To support customized Copilot development, Microsoft has made Azure AI Studio widely available, enabling users to create and deploy custom-built Copilots responsibly. Azure AI Studio’s dual developmental approach offers a user-friendly interface that caters to both graphical and code-first preferences. With the latest AI tools at their disposal, developers can test, connect, and validate their Copilot creations using their own data within a secure trial environment.
Some of Azure AI Studio’s many key features include:
- Models as a Service (MaaS): This anticipated service enables developers to access powerful AI models without having to set up complex systems or dedicated virtual machines. MaaS provides a library of AI models, facilitating the creation of smarter applications.
- Multimodal models: Azure AI Studio supports multimodal models, enabling it to handle diverse data types such as text and images. This capability makes it feasible to develop interactive and versatile AI applications.
- Monitoring capabilities: Azure AI Studio offers robust monitoring features, allowing organizations to track the progress and performance of their AI applications. This ensures that users are kept informed of trends, quality, and operational metrics for apps in production.
Extensibility features for Copilot
Microsoft has bolstered the integration capabilities of Copilot through seamless extension, customization, and amplification, with a broader range of applications and services. This extensibility means that developers can now customize Copilot's functionalities to better fit specific workflows, enhancing productivity across a variety of domains.
Let’s look at how those outcomes are achieved:
- Copilot extensions: Copilot extensions are achieved through connectors, plugins, and custom Copilots. These methods allow you to customize and enhance Copilot experiences by integrating specific data and processes to tailor solutions. These extensions can be developed using low-code in Copilot Studio or pro-code in Visual Studio Code with the Teams Toolkit extension.
- Develop your own Copilot extensions: Alternatively, users can publish their own Copilot extensions using either Copilot Studio or Azure AI Studio, and then integrate them as extensions using the Teams Toolkit in Visual Studio Code.
Copilots “your way”: Two methods will be introduced for creating your own Copilots and publishing them as Copilot extensions. For a managed stack, users can opt to create declarative Copilots, by specifying workflow instructions, actions, knowledge, and triggers. Alternatively, using a custom stack to develop custom-engine Copilots will allow users to introduce their own foundational models, as well as orchestrate and host additional functionalities for a tailored experience.
Declarative Copilots are ideal for a variety of specialized scenarios, specific data sources, or targeted roles within your organization. Custom-engine Copilots, powered by the Teams AI Library, and built using the Teams Toolkit for Visual Studio Code, offer more control over the user experience, support specific language models, and can be published to Microsoft stores as SaaS solutions.
- Copilot connectors: A third option for enhancing Copilot cohesion is through connectors, which enable Copilot to access relevant data for queries and actions. Copilot connectors can be built using ready-made connectors for various data sources, apps, and workflows, or by creating customized connectors with code for ongoing innovation.
- Copilot handoffs: Another pivotal feature is Copilot handoffs, which handle the process of transitioning a conversation from Copilot to another bot service that requires specialized knowledge or actions. This can include such things as IT support queries or product inquiries, with a seamless switch to custom Copilots.
A unified interface in Microsoft Teams
One of the most prominent features announced during the Microsoft Build conference was the ability to access various Copilots from a single interface within Microsoft Teams. This centralization feature is designed to streamline the user experience, allowing team members to leverage different AI assistants without needing to switch applications. By integrating Copilot directly into Teams, Microsoft aims to foster a more collaborative and efficient work environment.
There are several upcoming features for Copilot in Teams that will enhance and streamline user experiences:
- Message extension plugins with actions: Users who have the Teams app featuring a message extension can integrate this functionality into a plugin for Copilot. This will allow message extension plugins to execute actions on external systems on behalf of the user.
- @ mention plugins: Users will soon be able to access a platform capability that enables plugin activation with a simple mention in Teams messages. This integration will optimize the use of targeted knowledge for users, ensuring seamless workflow continuity.