Introduction to Amazon Quick Suite - Part I | Amazon Web Services

Key Takeaways

Hello everyone. This is Rajesh Pichani and I'm a solutions architect at Amazon Web Services. In this session, we'll go over some of the capabilities of Amazon Quick Suite in action, shall we? As always in our session, we'll start with a quick overview of the topic and this is going to be a quick site today and then we'll switch into the console to go over the capabilities of uh research agents and also spaces as a starting for this session and finally we will conclude with the references. Amazon Quick Suite is an agentic AI powered digital workspace that provides business users with a set of agentic teammates that can quickly answer questions at work and turn those answers into actions. That being said, Quick See, as you're seeing in the screen, has several capabilities and today we are going to talk about some of them using this workshop referenced here. The QR code for this workshop is highlighted in the top right for your access. We'll be covering spaces, research agent, and actions for this session. Let's go over to the console to see this in action. This is a link to the workshop that I referenced in the slide before. As you will be going through, you will start with this workshop materials. Download this file that consists of several subfiles underneath the zip file. We will be using this as we go along in executing this session. On the right side I have the console here. This is the homepage of Amazon quick suite. Depending upon what are some of the activity that you have done recently it might show up here but not necessarily the same as we are seeing right now. Okay. We will start with the unified chat very high level for now because as we will go through these other ones we will be interacting with the chat quite a bit here. Okay. So chat out of the box you will see this my assistant as a default one and this is uh going to be an conversational interface for uh your our users right to access this uh as agent we can click on this top right open chat or even this one as well which opens up the page just like this for starting our interaction here you will see the default assistant um picked up but as you keep building up more uh um agents you will see them uh added here that can be selected as well. Okay. Down below here we will have some um ability to choose general knowledge or focus on several um uh other applications or spaces that we will be uh interacting. These ones uh this enables a web search to go and uh crawl the web to get some answers for us. This is a file upload and this is a flow that we will touch upon later. Okay, that's a very high level overview of the default assistant. As I mentioned, when we get into spaces and flows, we will go into more details. In the spaces, think of this as a logical container for any particular use case or a domain or a team that you want to have access. right here in this uh workshop. This is taking an example of a HR department and that's where we will be using some of the materials that we have downloaded earlier. To create a space, click on creating the space. This has different um components to it. So the way you can have this uh spaces organized is through files either through simple file uploads or quicksite dashboards or knowledge bases actions and topics. For this exercise we will start with the simplest one file uploads. So in the interest of time I have pre-created all of this prior to the recording. So you are seeing here two activities HR operations and policies. If I click one of them, you're seeing these two files which were uh downloaded the zip file and then extracted out of it. Two files related to employee onboarding are in HR operations and several other files related to company policies or in HR company policy spaces. Okay. And assuming that this information contains very specific to your organization, um once this whole setup has been done, any user who is interacting with the chat agent can get answers pertaining to that particular organization. Right? Now that we have this spaces created, we can go and start that agent. And here the very first time you will not have this uh HR company policies or operations that was done by adding them right through this. Once we had the spaces, we then add here and that becomes a part of a knowledge search in responding to the questions. So here were a couple of questions that were asked after the spaces was created. So the first question was this company diversity and inclusion policy. The chat responded and along with the response we also see the sources being referenced back from the employee handbook PDF file when asked about the company's policy um policies leave policies which are very specific to an organization right so this information is pulled up from the leave policy PDF So we know the source [snorts] of this information is coming from this document. And likewise as we continue to ask our questions, the sources keep changing depending upon the question. The agent intelligently knows that uh this is a document that contains this answer and picks it up and responds. Okay. So that's the spaces part. Let me close this. And that's what has been outlined in this um workshop. The next one is research agent. Let me walk through the what I have done before. Say a scenario that you want to research on a particular topic or a um or a domain or a company or whatever the case may be that you want to dive deep into. In a traditional sense, you will have to scour through information from various sources throughout the web, collocate them and consolidate and present it. And this is a very laborious process and that's what research agent tries to um u help maximize and get this information in um relatively faster time. Okay. In this example work as outlined in this workshop. Let's go over this that was created. You can do this by also starting a research. Let me go over the steps here. When you click on new research, you specify the research objective, what you would like to see and um um the instructions to it. And then you will also provide um where can this agent go and uh do the research and come back with a report right? So it can go and do um user web and if needed you can also enable the with the files and also you can add from other resources within quick suite as well and finally third party. So here are the common uh sources of information that this research agent can use this in providing you a report. Okay, being said, I click on this. If you when you go through this workshop, you will see that um in the research objective field, you are actually putting this information. This the best practice for remote policies. This is what you provide in the research um objective field. The web search is automatically enabled. And in the quick suite assets, we also have enabled the space the policies space that was created in the previous uh exercise. And then finally, we start with the create plan and then start the research. It'll take approximately an hour or so. Um again it varies depending upon the complexity and scope. However, at the end of this uh you don't need to wait in the same screen you can close the window and come back and uh there'll be an email notification to the user telling that the information is completed. Once it has been completed so you come back and have a look at it about the research. As you are seeing here um this particular request completed with this comprehensive document of all the information based on the objective that we were we have given. If for some reason that you want to deep dive you can you can provide some comments add some revisions to it as and when you make the changes uh this gets version control so to speak um version 1 2 3 etc. And finally either you can do a different version of summarization either an executive summary or or different custom summary and then you can share it with the rest of your team. Right? So that's a quick uh overview of this research agent and finally the actions in this session. So that is done through uh integrations. Here there are two examples given in the workshop connecting to APIs and then connecting to MCP servers. And let me walk you through the detail. For this purpose some information is been um already provided as a part of the workshop. So when you set up this workshop, there are two buckets. And in this workshop artifact bucket, you will see these two files. You will have to download and keep it available. The JSON file and then the workshop details text. So this text file contains two information. One is for the API configuration um which is for the connecting to the APIs and the other one is for MCP configuration. Here are some of the um authorization uh configuration that is needed. So keep this handy. So the way to set this up is in the integrations actions. We'll start with the first one. So, uh let's go over the open API specification, import the schema. You will provide the name, description, a public network as a connection type. In this user authentication you will refer back to this file to have all the configuration the base URL client ID secret etc etc. We'll provide all of them in this corresponding uh entries and then create the integration. Once it is successfully integrated you will see them down below in the existing actions workshop APIs. In this um through this API you can have different action you can create as outlined here. You can test couple of them. Testing um list employees for example permit. Okay. Here is the response that um provided by the output of the API and in the upcoming sessions we will be using other activities or other APIs as well. Okay. So this will be coming in handy at a later point of time. Once the workshop AP has been uh integrated successfully, we can also do this similar one for the model context protocol. Um except that the details needed for different integrations might vary. The concept remains the same. It starts with providing the name and the description. In this case would be uh MCP server endpoint. For this you go to the text file and this would be the MCP server endpoint and when you go to the next screen in the authenticate you will be prompted with the rest of the information client ID secret etc. Once that has been successfully done and integrated, we will be using this uh credentials for testing it out. Okay, since I have already went through this configuration, um I'm showing the pro process here. After successful one, again we can do the test action API here. and then submit. This MCP server um is we are trying to get the latest uh AWS news and here are the list of the news here. Again this is just a testing UI and in a short while we will show you how it can be uh extracted in a natural language way. Okay. So we have this integration done. The way to um access this information is again um I'm just going through the previous history and showing you the results how it looks like. So when we ask what's the new what's new in AWS this week here it uses the MCP um connection and then provides us the response here and just like the previous example when we click on source it knows that we have to use the uh integration the MCP integration and get the response back and also this this is exactly what's been outlined in this one when when I use this particular question here for example you're seeing this information here identified workshop APS now based the question. The agent knows that it needs to get this information from the workshop AP list um openings, job listings and it is responding back here. And we can also uh give a different example. It will still use the workshop API but then a different tool here create time of request based on the question that we have posted. This API needs different set of information here and uh we can take action by creating a submit button. Okay. So this is how you can start having u turning into answers into actions as well. And going back to the workshop um this is the MCP server that we have um provided. What's new in AWS this week? This is what we asked by local group. So in summary, the chat agent that we are using, we are redirecting our responses to focus on the spaces information which contain a couple of files and we can also integrate with different forms of with different forms of integration. One is through uh APIs, another one through MCP servers. Depending upon the question asked, the agent intelligently redirects a question to the appropriate connection integration and then provides a response back to us. And again depending upon the question as well, we can further take the uh action as we saw just a while ago. Okay. So just to summarize um the the components that we discussed today, we started seeing spaces as a logical component to have the files. We also saw the research agent capabilities and then uh some examples in action. Obviously there are more to it and uh however in this workshop we are following the u the steps to interact with the APIs as well as the MCP servers. Okay. And in the subsequent one we will go into flows and other components as well. Going back to the slides, we will pretty soon come back with a part two on the flows custom chat agents and the knowledge bases. To conclude, um this is a link for the quick suite documentation. For those who would like to have much more hands-on and also stay connected with the latest and the greatest news about quick suite, please refer to this documentation and I hope you found this session useful. Thank you very much and happy building on AWS. Bye.