Unlocking the potential of Microsoft Copilot and Copilot Studio
This blog has been expertly written by Chris Burns, Technical Architect at Nasstar.
Copilot and Copilot studio is everywhere and is fast becoming Microsoft’s new favourite child. At Microsoft Ignite 2024, the company unveiled significant enhancements to Copilot Studio. Notably for us, its integration with Azure AI Foundry and the introduction of autonomous agents. These developments are poised to revolutionise how organisations leverage artificial intelligence (AI) to streamline operations and foster innovation.
The evolution of Copilot Studio
Copilot Studio has evolved into a comprehensive platform, enabling users to create and manage AI agents with enhanced autonomy and intelligence. This progression allows businesses to automate intricate processes, thereby boosting operational efficiency and reducing manual workloads. The platform's user-friendly interface facilitates the development of AI solutions tailored to specific organisational needs, making advanced AI capabilities more accessible across various sectors.
Historically, Copilot Studio was primarily a chat-driven interface. The solution was always response-driven from an initial prompt from the user. In cheesy vernacular, the AI always had to be prompted to do something. Nasstar’s AI & Automation team have been building these agents for a while outside of Copilot Studio, and our development team are working on the feasibility of building “Multi-Agent workloads” on the Copilot platform.
Chris Burns, Technical Architect at Nasstar, said: "I have been working with Copilot Studio since the days of PVA and the pace of development has been outstanding, which I suppose is understandable based on the investment that Microsoft made in OpenAI. I am excited to see how CoPilot studio is going to empower citizen developers to work with the new Gen AI features, especially as the barrier for entry for building modern apps is so much lower than it was with Power Apps.
"The future of computing interfaces is very much the prompt and the computer just understanding your desires and needs. Imagine a world where holiday requests no longer need you to fill in a multi-page app but allow us to ask the HR bot: "Book me the days off over Christmas next year and email me when it’s done"!"
The introduction of autonomous agents
A pivotal advancement for Microsoft is the introduction of autonomous agents within Copilot Studio. These agents are designed to operate independently, monitoring events, making informed decisions, and executing actions without continuous human intervention.
For instance, an autonomous agent can handle customer returns by assessing requests, updating inventory systems, and notifying customers autonomously. This capability not only reduces manual effort but also enhances process efficiency and accuracy. The social aspect aside, these agents are an exciting evolution of the Copilot studio platform.
Chris said: “I love how this technology is aiming to make our lives easier. Autonomous agents grounded in your data is going to allow some funky productivity gains, not only for the host company but for employees too. Once this concept is understood, I think employees are going to find some amazing ways to speed up their work. Getting over societal change and luddite views is going to need some inflection on “What Work Means?” though.”
When first mooted, they appeared to be nothing more than a Power Automate flow that we already had access to. However, what transpired was that while the triggers were the same, the knowledge and actions were where they differed.
No longer did we have to create a flow of “if this then that”. We gave the agents the connectors, such as Email, Teams, and ServiceNow and told the agent what each connector’s capabilities are – “The days of skipping over the description box has now long gone” – we provide the agent with the knowledge it needs for the process, such as company refund policies.
The agent then uses all this input to decide which API connector to call and what data to provide to it and then listens for the output from said connector and decides on the next action to take. All this without being told specifically but guided by natural language input.
Practical application: Autonomous agents in supply chain management
Consider a scenario in supply chain management where an autonomous agent oversees inventory levels. The agent can predict stock shortages, automatically place orders with suppliers, and arrange for timely deliveries. By autonomously managing these tasks, the agent ensures a seamless supply chain operation, minimising delays and preventing stockouts. This application exemplifies how autonomous agents can optimise complex processes, leading to cost savings and improved service levels.
An example from Microsoft
https://www.youtube.com/embed/qxMi-tLg4MA?feature=oembed
Seamless integration with Azure AI Foundry
The integration with Azure AI Foundry amplifies Copilot Studio's capabilities. Users now have access to an extensive array of AI models and tools, including the prebuilt and custom-built Azure AI models. This integration empowers organisations to develop sophisticated AI solutions that address unique business challenges, encouraging innovation and enhancing competitiveness.
The collaboration between Copilot Studio and Azure AI Foundry also ensures that AI agents can be deployed across multiple channels, such as Microsoft 365 Copilot, Microsoft Teams, and various third-party messaging platforms, extending their reach and utility.
As has always been the case, Copilot Studio is a low-code development platform. But sometimes that environment does not scratch the itch. Azure AI Foundry, formally Studio, contains a range of Azure cognitive services. These tools are the brains behind Copilot Studio and allow much more advanced solutions to be built in code.
By consolidating cognitive services into a single platform, Azure AI Foundry enables developers, AI engineers, and IT professionals to customise, host, run, and manage AI solutions with greater ease and confidence. This unified approach simplifies the development and management process, allowing all stakeholders to focus on driving innovation and achieving strategic goals.
The marriage between Copilot Studio and Azure AI Foundry means these powerful models and solutions built in Foundry can now be surfaced with ease into Copilot and Copilot studio.
Future prospects and industry adoption
The advancements in Copilot Studio, strengthened by its integration with Azure AI Foundry and the deployment of autonomous agents, lets organisations fully harness AI's potential. These tools facilitate the automation of complex tasks, enabling businesses to focus on strategic initiatives and innovation. Industry leaders are recognising the transformative impact of such technologies. For example, companies like Pets at Home have utilised Copilot Studio to create autonomous agents for specific business workflows, demonstrating the practical benefits and scalability of these solutions.
Microsoft's enhancements to Copilot Studio, in conjunction with Azure AI Foundry, mark a significant milestone in the evolution of AI-driven business solutions. The introduction of autonomous agents offers a glimpse into a future where AI not only assists but actively manages and optimises business processes, paving the way for unprecedented levels of efficiency and innovation.
Speak to our team today to find out how Microsoft's latest technological advancements can help your organisation.
FAQs
How does Microsoft Copilot Studio improve developer productivity?
Microsoft Copilot Studio is not designed to improve developer productivity, but to support citizen developers, which is really the entire workforce.
Improving productivity means different things for different people. In the past, autonomations or applications were built to solve specific problems. Today, with copilot autonomous agents, time spent on repetitive tasks is low hanging fruit. Almost every employee can think of something in their day-to-day work that is repetitive, monotonous and can be automated with AI. Should that be reading an email and creating an order or scanning a document and creating an Outlook calendar entry.
Is Microsoft Copilot Studio compatible with Visual Studio Code?
Copilot Studio is not compatible with Visual Studio Code, but GitHub Copilot is another product from Microsoft which has deep ties with Visual Studio Code.
Can Microsoft Copilot Studio be customised for specific coding languages?
Microsoft Copilot Studio has no customisations for specific coding languages as such, but the LLM which underpins Copilot can create code from a multitude of languages such as JavaScript, Python, or C#.
Are there any alternatives to Microsoft Copilot Studio?
Many applications are building AI bots in their products as alternatives to Microsoft Copilot Studio, such as Sage Copilot, for accountancy. But Microsoft Copilot Studio has advantages in that it has native interaction with M365 services. No other chatbot provider is that embedded in M365, and by extension your business data.
