Azure AI Foundry Agent Service: Transform agentic workflows

Key Takeaways

You can now automate the unautomatable. But Azure AI foundry agent service is generally available. That's huge news. Woo. This is a great time to get started. How are we using agents to help on our own processes? So there's a a lot of emails that we receive of like, hey, what about this functionality? I tried this. Here's a screenshot. This is a real issue that we had the other day. A picture of that issue is a big error and they send it off. And typically what they would have to do is wait a few days for me to respond because I'm a very busy human. But instead what they were able to do is get a response from an agent. And this agent was able to do some work by automatically creating a ticket in ADO. An ADO in DevOps. Right. Okay. So now we're going to build an agent. Select our lovely agents tab that no longer says preview. Hooray. Um, and I could click on this new agent button and type some stuff into this form, but I am not going to do that. This is built for prodevelopers like yourself. You want to write code. And we can see here constants that actually map to all of those fields that we saw in that form. So, what do they have? An ID. An ID. Yeah. A name. A name. The instructions. The instructions. And the tools. Tools. Okay. And we can just give this to our SDK, our APIs in a few lines of code. And I'm actually going to hit the run button and create that agent inside of our agent service. All the control. All the control. Can see that customer support agents. Woohoo. Yep. Try it out in the playground, which you can also do in the VS Code extension. Think about them as microservices. When you build your end-to-end system and something changes, you can easily fine-tune, you can easily tune that like specific node of your solution without having to rebuild everything else.