I'm having technical difficulties. So the first thing we're going to show here is just we can have different setups since each of the agents that Amon was showing you, we can give them not only different tasks but also different LMs. So the one that we're going to be showing here is using Claude Son at 3.7 as the supervisor and the agents below that are all going to be using GPT-4, but we can switch to using like a Mistral local with a Cohera.

For this demonstration, you're looking at Son at 3.7 and OpenAI's LLM. So this would be a typical question that a renewals manager might ask, show me the top five deals aggregated by ATR. Also, I want to point out that ATR may not be understood by the LLM, and we'll show you later on how it understands what ATR is within the context of this question.

For all the customers, along with their renewals risk severity level sentiment classification, an adoption score, and a tabular format. So this question is a good one because it will utilize all of the agents that we were just showing. And so we're going to hit send on that. I'm probably going too fast.

And so while the agents are working on answering that question, we're just going to switch over again to the infrastructure that we have here. And if we look at what we have as far as the agents and then look at the studio version of that, we can see that the supervisor is going to create an action plan.

Once it's created, the action plan is then going to call one or several of these agents. And then once the agents are complete, it will go back to the supervisor, which will synthesize the results as the final answer for the user. And so if we look at each one of the agents individually, we can see that they're all set up similar, where they're going to handle the input.

So it's going to format the input properly. And then we have a react style for each one of these. And so they're all pretty much set up the same, except they have access to different tables. Their prompts are different. So they're being set up to specialize in the tasks that they're meant to do.

And then Aman also mentioned the discovery agent, which I'm not sure we're going to have time in the presentation to go over. But the discovery agent is there for those questions that come in that are very vague. And we don't quite-- the LLN doesn't quite understand what the user wants.

But it's given access to a semantic model, which essentially is all the metadata for the database behind it so that it can hold a conversation and extract from the user what they're actually looking for. And we know we can answer that question because the conversation is based on the metadata for the tables behind the answer for that.

And so we're back now to that question that I asked, and we get a nice rich output for that. And so a renewals manager could use this to start looking for things that they should work on. But before we get back into the business use case, we're going to take a little behind-the-scenes look.

And so a lot of this, you saw Lance show earlier. And so we're going to just look at the traces for what was going on. So here we can see the AI assistant, which essentially is the supervisor, is involved. And then hands out the tasks to the renewals, the sentiment, and the adoption agent.

And so if we look, again, what's going on at the supervisor level, the first thing that's going to happen-- first we can see that within Linesmith, we can see all of the different nodes that we're executing within the graph here. And as we dive in deeper, the first thing it's going to do is categorize the question.

Does this question need an agent? Is it something that's unrelated? Or does it need discovery because it's a vague question? So in this case, it identified that agents are required. It gives the reason for that. The first thing we do is each agent has also access to a vector store.

So the first thing we're going to see is do we have any examples of the question that I'm being asked. And if we do, we're going to use those examples. In this case, there was no examples. So the LLM is going to try to dissect that question and decide which agents are needed.

And so in this case, it decided that the question requires both the renewals, the sentiment, and the adoption agent. And breaks up the question into individual tasks that those agents have to perform. And so if we now go in and we look at what the renewals agent does, again, we can see that the nodes within the graph are well represented within Linesmith.

And so now we're going to go down to the end here. We're going to start walking through what the renewals agent would do. So the first thing we see are the tools that the renewals agent has access to. So it has access to four SQL tools where it can list the tables.

It can create a query. It can get the schema of the table. And there's a query checker. We also have search examples because for the renewals agent, it needs to do some research on the internet for some of the questions that they may ask. As well as, well actually, search examples from the vector store and the web searches for anything that needs to go out on the web floor.

Then we see the system prompt. Oh my gosh. Sorry. A lot of technical things. Then we see the task that it was assigned by the supervisor. We then see that it's going to search for examples. In this case, it does find an example. As part of that example, we give it steps that it could do.

In this case, this is a very simple example. We just literally give it the English steps that it needs to follow, as well as an example output that we would like the output to be. It then goes and lists the tables that it has access to. It then goes and gets the schema of those tables.

Now as part of the schema, we also have been leveraging, Snowflake has something known as a cortex semantic model. So we're not using their LLMs, but we are leveraging their data model for doing semantic modeling, which essentially is metadata for the tables. So here we can see we have the column names, but we also have sample values that are in that column, synonyms that a user may say, you know, client ID, which we can then map back to the column name of the cat view ID.

There's other information here, such as for this total customers is, I mean, this one is a simple one. It's just the count. But in some cases, things like filters or calculations are in here. Here's one, for instance, where high-risk deal count. It probably wouldn't know that offhand how to do that, but here we're telling it how to calculate that.

And so this metadata helps a lot to get the accuracy of the LLM up as people are asking questions. And I'm going to go through this a little bit quicker since I'm running out of time. But essentially, we'll create a query based on that, get an answer, and then synthesize the result.

Here, notice that this output matches the example output that we were showing up further, so that's part of that example. And each one of these agents is then going to do that same procedure, so I'm just going to go through this quick. And then once all three of those agents have done their job, it goes back to the supervisor, who's going to take that output and synthesize a final answer, which is then displayed back to the user.

And I'll just quickly go through this so the user can then use this information. So here, he's looking for high-risk opportunities. It's also highlighted as key insights, as well as it's a recommended action for him to follow for this customer since their renewals is at high risk. And unfortunately, I'm not going to be able to make it through the rest of the demo, and I'm going to hand it back to Amon.

So please come see us at the booth if you'd like to see the rest of the demo. Thank you.