Microsoft And Openai Just Unleashed AI Agents - (Copilot Agents + Demo)

Today, Microsoft put the AI industry on hold. They announced their new autonomous AI agent capabilities across Copilot Studio and Dynamics 365 to help scale the impact of every individual. This is Satya Nadella, the CEO of Microsoft, tweeting that today they're announcing exactly what agents are going to look like. And in this detailed video, I'll show you everything you need to know about the coming rise of Microsoft's agent and and how they're going to impact the future of your work.

And lastly, we will show you how you can extend Copilot with agents. Copilot is becoming this new enterprise orchestration layer. Yesterday's bespoke interactions with siloed business applications will just simply become Copilot agents. I cannot wait to see what you do with all these new capabilities.

So here we have it. Copilot is your AI assistant. It works for you and Copilot Studio enables you to easily create and manage and connect agents. To Copilot, it says, think of agents as the new apps for AI powered world. Every organization will have a constellation of agents ranging from simple prompt and response to fully autonomous. I think we need to truly understand that statement. They're currently stating that every company in every organization is going to be transformed. Imagine having an army, a workforce, a constellation, as Microsoft put it, of agents ranging from simple prompts to fully autonomous. And that's one of the things we're going to dive deep into this video because fully autonomous agents are pretty next level. They can complete a series of tasks over a longer time horizon and execute more work, saving you more time. It continues to state that they will work behalf. It continues to state that they will work on behalf of an individual team or function to execute and orchestrate business processes. Copilot is how you'll interact with these agents and they'll do everything from accelerating lead generation and processing sales orders to automating your supply chain. Also goes on to talk about empowering more customers to build autonomous agents in Copilot Studio. It says earlier this year we announced a host of new powerful capabilities in Copilot Studio, including the ability to create autonomous agents. And next month, this is where the key thing is, because this is the date next month these capabilities are shifting from private to public preview, allowing more customers to reimagine critical business processes with AI. And it says here, which I've highlighted, agents draw on the context of your work data in Microsoft 365 Graph Systems of Record, dataverse and fabric and can support everything from your IT help desk to employee onboarding and act as a personal concierge for sales and service. So let's not focus on the mumbo jumbo. Let's actually take a look at what these agents are going to do and what kinds of agents we're actually going to be getting and working with. So on the article it says that today we're announcing new agentic capabilities that will accelerate these gains and bring AI first business processes to every organization. First, the ability to create autonomous agents with Copilot Studio will be in preview next month. And Second, they're introducing 10 new autonomous agents in Dynamics 365 to build capacity for every sales, service, finance and supply chain team. So it seems Microsoft is ramping up their ability to have agents that are doing real tasks in the real world. So this is where Microsoft talks about autonomous triggers. So this is basically how agents are going to work. So it says that agents can automatically respond to signals across your business and and initiate tasks. They can be configured to react to events or triggers without human input that instead originate from various tools, system or databases, or even scheduled to run hourly, daily, weekly or monthly. So you can see here that this is quite like automations, but I'm guessing that this will be an advanced automation. So you can have a trigger set so that when something in your business occurs somewhere, you can have the agent that picks up on that and then.

Of course reacts to that.

In this example you can see that this agent is a supply chain multi agent and it is being configured to have certain triggers that could help out the business. Next is where we've got the dynamic agent plan. So this is where we can actually see exactly how these agents work. So it says each business process can have different paths since agents create dynamic paths on the fly to handle and complete tasks. And users can view the underlying logic for each of the agent's paths, which include key details, steps and systems involved. This provides insight into why the agent.

Chose a particular method, its decision process.

And context, along with detailed steps including variables and outputs which are crucial for debugging. So overall what we can see here is the agent plan and luckily we can actually see what these agents are actually planning, understanding exactly what they planned and of course changing these plans to get certain results. I think it will be interesting to see what kind of plans agents choose as it says that agents can create dynamic plans on the fly to handle certain tasks. Then of course we do have the.

Activity overview which essentially shows you your previous agent run.

So in the article it says that trust, dependability and precision are at the.

Core of establishing agent autonomy in organizations.

The true strength of these agents is.

Their ability to monitor activities across real world scenarios, which in turn enhances business outcomes. And to further transparency and accountability, we're launching an activity tab that enables makers to access a complete log of past agent runs, including progress tracking, identifications of blocks, trend analysis and review earlier decisions. So in this area, what you're going to be able to see is, is everything that your agent has done, if it is completed and the apps that they use, you can see that on the right hand channel the channels essentially which things these agents use in order to complete that task. So this is going to be something that is really transparent. So for those of you who might.

Be worried about agents doing strange things.

Behind your back and running off strange parts of the Internet, that is going.

To be something that won't happen.

Now there's no surprise that this is going to be using the most advanced OpenAI 01 model. It says that Copilot Studio agents use the latest models, including the OpenAI 01 series, which is being used in the limited private view for autonomous agents. These models are optimized for advanced reasoning tasks, enhancing the ability to solve complex problems.

And if you aren't familiar with the.

OpenAI 01 series, those are the models that are quite different to your standard GPT4. Oh, the OpenAI 01 series is pretty much trained exclusively for reasoning, which makes them perfect for these scenarios. One of the things that most people.

Didn'T realize is that whilst yes, you.

Can use OpenAI01 for a variety of different use cases, one of the biggest use cases that we're going to be seeing on the backend is of course autonomous agents. And with the advanced reasoning, I can't imagine that these kinds of agents are going to fail. And being able to plan, being able to execute certain tasks and have it all transparent in one area is going to be extraordinarily great for us who want to know exactly what is going on and of course for the reliability of exactly what we're trying to do. So this is where we have additional information for AI agents. Now, I think AI agents are a new territory, a new area, because it's quite uncharted. We haven't had really autonomous agents in any sort of capacity.

So this is something that is quite.

New, which means that we're going to have to ensure our safety. Security is top notch. So essentially Microsoft state that organizations today have what they describe as a wealth of tools at their disposal to ensure safety and security of their AI.

Systems.

Now, when they say protect sensitive data, they use enterprise grade measures designed to keep everything secure, which is basically what we can think of like a high tech vault, complete with multiple levels of encryption, which basically just scrambles your information so only those who should see it can see it. Now they also talk about data prevention measures, which is where they make sure that no data gets misplaced or falls into the wrong hands. And a responsible AI strategy to handle unique related AI challenges, like preventing prompt injection attacks, which are, as of recent papers, pretty easy to do. And these are essentially ways that people might try to trick an AI into giving away sensitive information. So the data used by these AI agents is linked to strict security controls that are overseen from a centralized admin center, basically mission control, where these rigorous authentication protocols ensure that only authorized people can access this sensitive data data. So the way that these AI agents behave is also going to be tightly controlled. Organizations can set specific guardrails to define what these AI agents are allowed to do, much like giving detailed instructions to a team member. And these instructions include what the AI can access, what actions it can take, and how it should respond in different scenarios. Now organizations can decide who, who gets to create these AI agents, who can share them and who can use them. And additionally, they can put policies in place, like labeling certain information as highly confidential so that the AI knows how to treat these different kinds of data. It's a bit like setting different levels of clearance, ensuring that sensitive information is handled appropriately. Now, managing these AI agents over time is of course going to be just as important as setting them up. And this is basically like where you're overseeing a project from start, start to finish. These organizations have a process governance that allows them to manage the AI agent's journey from creation all the way to retirement. And they can determine who can access the resources, decide how these resources are used, and of course, control when an AI agent is ready to be released for wider use. So beyond that, they have the ability to audit what the AI agents are actually doing and monitor their performance. So if these AI agents aren't doing well, they might get fired. Just kidding. They might make necessary adjustments to keep improving how they operate. And this ongoing monitoring helps ensure that everything is running smoothly and complies with the required regulation and standards. Overall, a comprehensive security picture for Microsoft, resulting in your data being secure. Considering the fact that we have these autonomous agents, I would hope so. Now, in terms of the examples of agents, Microsoft gives us an insight to the kinds of agents that are going to be available, are going to be created. So, of the 10 autonomous agents that Microsoft is offering, one of those is going to be a sales qualification agent. In a profession where time literally equals money, this agent enables sellers to focus their time on the highest priority sales opportunities, while the agent researches leads, helps prioritize opportunities, and guides customer outreach with personalized emails and responses. This is one of the first examples that seems to be focused on doing mainly the tedious tasks that many people.

Wouldn'T want to do.

Of course we have number two, which is a supplier communications agent. And this agent enables customers to optimize their supply chain and minimize costly disruptions by autonomously tracking supplier performance, detecting delays and responding accordingly, freeing procurement teams from time consuming manual monitoring, procurement teams from time consuming manual monitoring and firefighting. Once again, an agent doing the communication which seems like once again, in the future it's going to be a lot of AIs talking to a lot of other AIs. Now of course, we have the customer intent and customer knowledge management agents. A business gets one chance to make a first impression and these two agents are game changers for customer care teams. Facing high call volumes, talent shortages and heightened customer expectations. These agents work hand in hand with a customer service representative by learning how to resolve customer issues and autonomously adding knowledge based articles to scale best practices across the care team. So overall we can see here that a lot of these agents seem to be in some kind of support role. They don't seem to be taking on the majority of the work, but they will be doing a lot of many different subtasks that are vital to the business and but just time consuming. So these are the kinds of agents that Microsoft is going to be releasing.

I think we should take a look.

At what Microsoft have actually said regarding these agents.

Earlier this year they did the Wave.

Two presentation in which they gave us a really good showcase on what some of these agents are going to be. And I know I've already spoken about these agents in key detail, but seeing this presentation, this short snippet is going to give you guys the kinds of insights that you need to understand where these agents are going to head. And then after this I'll get into some key ways organizations already using these agents and taking advantage of them.

Introducing copilot agents Broadly speaking, agents automate and execute business processes working with or for humans. They come in all shapes and sizes. They can reason, remember, be trained and even know when to ask for help. Agents range from simple prompt and response to more advanced agents that automate repetitive tasks to sophisticated agents that are fully autonomous and can orchestrate other agents. Copilot agents do that and more. They bring the power of agents to you right in the flow of your work. They work for you in the background, are fully managed and orchestrated by copilot, and are incredibly easy to create. Copilot agents build capacity, enabling you to scale your team like never before. Take a look.

We'll start in Biz Chat where we can open a simple agent for a field service technician. Say you're on site and you get an error code you've never seen before. You built a field service agent for exactly this. It has access to all your product knowledge and can give you step by step instructions in real time. Now watch how simple it is for anyone to create the agent we just saw and add it to copilot. You fire up Agent Builder and describe what you want the agent to do. It needs to be able to reason over your product catalog and repair tickets to provide step by step troubleshooting instructions. Let's connect it to SharePoint so it has access to all the knowledge it needs. It's built in seconds so you can start using it immediately or edit and configure it to fit your needs. You can expand its knowledge base by tapping into resources like SharePoint sites or other IT approved data sources. You can even update its name or image to align with your company's branding. So let's do that. You can test the agent right here or just hit Create and you're done. And of course the agent respects your organization's Microsoft 365 and SharePoint security permissions. Now your Copilot agent is ready in Biz Chat and you can share it with your team. Over the years, SharePoint has become the most used solution in the enterprise for knowledge and business process. Today we're enabling you to turn every SharePoint site into an agent, moving far beyond natural language search so you can fully leverage the value in its vast knowledge repository. Let me show you. Let's say there's an issue that requires further research and follow up. Your team keeps all its customer records from deployment to maintenance reports on a SharePoint site. It's a wealth of valuable insights that's largely been locked away. Kind of hard to extract until now. You can create an agent with a single Click from any SharePoint site library or folder. An agent that deeply knows the information it was built on. It can answer questions about source material, reason over that material and take action just like a well informed teammate would. In just a few seconds, your copilot agent is ready to be used or shared with others. Let's add it to your team chat so everyone can work with the agent. Now all Those artifacts in SharePoint are immediately accessible and useful, so Copilot can reason over all the relevant content to find what your team needs. You can ention the agent as you would any other teammate, sharing new information and asking it questions and it will respond in real time. Let's take it a step further. Now we'll edit the agent we just built so we can take action on behalf of the team. First, open your agent in Copilot Studio. Here you can customize the knowledge an agent has as well as the actions it can take. You want it to be able to place an order for new parts with your procurement team. Let's connect it to Dynamics 365 Supply Chain Management and in just a few seconds you're ready to update the agent and bring it back to the teams chat where you're working on solving the customer's issue. Let's ask the agent to order the part it recommended. Your agent is on it asking you to confirm before moving forward. Once you give it the go ahead, it's done.

Now the crazy thing about this is that this isn't just speculative. Companies are already using these kinds of agents to already increase revenue. You can see here it says organizations like Clifford Chance McKinsey & Co. Pets at home and Thomson Reuters are already creating autonomous agents to increase revenue, reduce costs and scale impact. Pets at Home, the UK's leading pet care business, created an agent for its profit protection team to more efficiently compile cases for skilled human review which could have the potential to drive seven figure annual savings. And what's the craziest thing about all of this is that Microsoft aren't just all talk, they're actually using AI agents themselves at Microsoft. It says here at Microsoft we're using Copilot and agents to reimagine business process across every function while empowering employees to scale their impact. Using Copilot 1 sales team has achieved 9.4% higher revenue per seller and closed 20% more deals. And thanks to Copilot 1 team is resolving customer cases nearly 12% faster. And our marketing team is seeing a 21.5 increase in conversion rate on Azure.com with a custom agent designed to assist buyers. And in human resources, our employee self service agent is helping answer questions with 42% greater.

Accuracy.

Overall. This is absolutely crazy. A lot of the times we might see products released by companies, but very rarely do you see the company implementing those.

Themselves.