WEBVTT

NOTE Sentence-level transcript of https://www.youtube.com/watch?v=Y_p3y6zP5QE

NOTE One cue per sentence. Cue ids are the line anchors on /transcripts/Y_p3y6zP5QE.html. A cue ends where the next begins, or 2 s after its last word.

s1
00:00:01.309 --> 00:00:03.309
[music]

s2
00:00:12.520 --> 00:00:12.960
Hi everybody.

s3
00:00:12.960 --> 00:00:14.040
Thank you so much for coming.

s4
00:00:14.040 --> 00:00:23.520
My name is Jack Witherspoon and I am a DevEx engineer at Google DeepMind and today we're going to talk about managed or cloud agents.

s5
00:00:23.520 --> 00:00:30.360
So raise your hands here, who all has heard of either cloud or managed agents or has used them?

s6
00:00:30.360 --> 00:00:31.080
Okay.

s7
00:00:31.080 --> 00:00:34.240
I'm starting off on the right foot.

s8
00:00:34.400 --> 00:00:45.440
So you might have heard of them if you for those who haven't, cloud or managed agents are essentially a way of instead of running a harness and running coding agent on your own machine,

s9
00:00:45.440 --> 00:00:50.040
you can actually just run them in the cloud on a sandbox.

s10
00:00:50.040 --> 00:00:54.160
So let's first start by why the heck would you even want to do that?

s11
00:00:54.160 --> 00:00:58.200
Well, sandboxes are really powerful for a number of reasons.

s12
00:00:58.200 --> 00:01:00.640
Number one, they're not running on your own machine.

s13
00:01:00.640 --> 00:01:04.080
So you can, you know, spawn a cloud agent.

s14
00:01:04.080 --> 00:01:05.640
It can go and write code.

s15
00:01:05.640 --> 00:01:06.680
It can reason.

s16
00:01:06.680 --> 00:01:09.000
It can do research, search the web.

s17
00:01:09.000 --> 00:01:16.840
It can go ahead and write tests, run tests, verify and actually come back with verified kind of answers.

s18
00:01:16.840 --> 00:01:18.520
And that's a really strong thing.

s19
00:01:18.520 --> 00:01:24.160
If you've ever like tried to run verification either on your own laptop, you know, sometimes it steals the browser.

s20
00:01:24.160 --> 00:01:25.760
It starts to run things for you.

s21
00:01:25.760 --> 00:01:27.720
You start getting pop-ups and things like that.

s22
00:01:27.720 --> 00:01:32.640
So running this all in the cloud takes away that pain.

s23
00:01:32.640 --> 00:01:36.160
And the other reason is that your laptop doesn't scale.

s24
00:01:36.160 --> 00:01:44.640
You might start to have specific setups like agent skills or you maybe have customized your personal coding agent on your machine.

s25
00:01:44.640 --> 00:01:50.320
But then how do you go and share that with a colleague or how do you integrate that into your app, right?

s26
00:01:50.320 --> 00:01:57.840
You don't want your app that you're deploying to customers to be running with your own kind of subscription or things like that, right?

s27
00:01:57.840 --> 00:02:01.160
So, we want to be able to scale this up and doing so.

s28
00:02:01.160 --> 00:02:05.840
And managed agents or cloud agents are a great great way to do that.

s29
00:02:05.840 --> 00:02:08.240
So close, the sign almost stayed up.

s30
00:02:08.240 --> 00:02:09.036
Don't worry about it.

s31
00:02:09.036 --> 00:02:11.036
[laughter]

s32
00:02:12.840 --> 00:02:20.360
So, when it comes to kind of managed or cloud agents, you know, to do this yourself is quite a tedious process.

s33
00:02:20.360 --> 00:02:23.800
You know, you have to actually spin up a container or a VM.

s34
00:02:23.800 --> 00:02:27.320
You have to load things onto that, manage it.

s35
00:02:27.320 --> 00:02:33.200
And then you actually have to do harness engineering, you know, figure out kind of build your own harness within the sandbox.

s36
00:02:33.200 --> 00:02:39.680
And then let's not even start to talk about networking or setting up firewall rules or all the security stuff.

s37
00:02:39.800 --> 00:02:48.959
So, that's where kind of managed agents come in and whether it's Gemini managed agents, cloud managed agents, we'll be talking about Gemini managed agents today, which is our offering

s38
00:02:48.959 --> 00:02:51.680
in the Gemini API at DeepMind.

s39
00:02:51.680 --> 00:02:57.160
And really now it boils down to just being a single API call.

s40
00:02:57.160 --> 00:03:00.320
So, that is really the beauty of managed agents.

s41
00:03:00.320 --> 00:03:04.000
With a single API call, you're getting all the things we talked about.

s42
00:03:04.000 --> 00:03:13.000
So, you're getting the sandbox, you're getting the anti-gravity harness built out of the box that is in the sandbox, and it can go ahead and do a lot of powerful things.

s43
00:03:13.000 --> 00:03:24.400
You don't have to worry about managing any of the infrastructure or the scale as you need in your application more agents, you can just do more API calls.

s44
00:03:26.640 --> 00:03:28.360
Actually, we'll go back for a sec.

s45
00:03:28.360 --> 00:03:31.640
So, you can see here, this is our interactions API.

s46
00:03:31.640 --> 00:03:37.480
It's our kind of new agent first version of our SDKs and client libraries.

s47
00:03:37.480 --> 00:03:47.480
So, previously we had kind of models, you would, you know, do an input, get an output, but that didn't really scale to agents in the new world we live in.

s48
00:03:47.480 --> 00:03:51.880
So, we launched a new kind of SDK uh, covering interactions.

s49
00:03:51.880 --> 00:03:55.680
So, interactions are much more kind of mirroring what agents are doing.

s50
00:03:55.680 --> 00:04:04.240
So, it can be multiple steps, you know, when a when you do an agent call, it can go ahead and write files, call those files, call different MCP tools.

s51
00:04:04.240 --> 00:04:13.000
Um, so really this new interactions API that we launched, uh, and went GA earlier this year, is now our kind of home for managed agents.

s52
00:04:13.000 --> 00:04:16.760
So, you just call the agent instead of having a model, you have an agent.

s53
00:04:16.760 --> 00:04:19.359
So, we have our base anti-gravity agent.

s54
00:04:19.359 --> 00:04:21.640
And then you set environment equals remote.

s55
00:04:21.640 --> 00:04:27.440
So, that's that key thing that's now telling it, "Okay, we want a managed agent that runs in the cloud."

s56
00:04:27.440 --> 00:04:31.600
And then you give it kind of regular input as you're used to.

s57
00:04:32.160 --> 00:04:36.120
So, here's kind of how the infrastructure actually looks under the hood.

s58
00:04:36.120 --> 00:04:39.760
So, when you call a managed agent, you kind of have that input that we saw.

s59
00:04:39.760 --> 00:04:50.760
And then it calls the Gemini API, which has a bunch of, you know, built-in server-side tools like, um, searching the web through Google Search or file system tools to read and write files.

s60
00:04:50.760 --> 00:04:56.680
Um, and you can also add your own custom functions or MCP tools as well.

s61
00:04:56.680 --> 00:04:58.200
And then we have the sandbox.

s62
00:04:58.200 --> 00:05:01.360
So, that's actually running on kind of Google Cloud infrastructure.

s63
00:05:01.360 --> 00:05:08.600
Um, so when you call the Gemini API to get a managed agent, it goes ahead and actually sets up and spins up that sandbox for you.

s64
00:05:08.600 --> 00:05:11.520
And you can go ahead and actually upload files to the sandbox.

s65
00:05:11.520 --> 00:05:21.560
So, if you have agent skills or agents.md, or you want to load a GitHub repository on startup, uh, we allow you to do all of that cool stuff.

s66
00:05:22.919 --> 00:05:26.960
So, here are kind of the building blocks that make up a managed agent.

s67
00:05:26.960 --> 00:05:29.560
These are kind of the different core pillars.

s68
00:05:29.560 --> 00:05:32.880
Um, so you've got the harness itself, and that's the anti-gravity harness.

s69
00:05:32.880 --> 00:05:39.040
It's the exact same harness that's running, uh, in our anti-gravity suite of products over at Google.

s70
00:05:39.040 --> 00:05:41.600
So, that's what's kind of figuring out the tool calls.

s71
00:05:41.600 --> 00:05:43.560
It has the built-in tools.

s72
00:05:43.560 --> 00:05:45.200
Um, and then we have the environment.

s73
00:05:45.200 --> 00:05:57.120
We'll talk a bit more about how the environment is kind of its own entity, um, which allows you to do really cool kind of multi-agent workflows by having different agents pointed at the same sandbox so they can write files.

s74
00:05:57.120 --> 00:06:02.840
A different agent can pick that same environment up later and actually build upon those files.

s75
00:06:02.840 --> 00:06:15.800
And then we've got skills we talked about so you can upload your skills to the sandbox just as you would have potentially like playbooks or different ways of doing things at your company or on your personal agent, you can upload those to the managed agent as well.

s76
00:06:15.800 --> 00:06:17.000
And then we have credentials.

s77
00:06:17.000 --> 00:06:29.680
Credentials is a new piece of managed agent which essentially lets us inject either kind of tokens or your MCP OAuth into the managed agent sandbox so you can access your different services.

s78
00:06:29.680 --> 00:06:34.000
And then we also have triggers which allow you to actually schedule a managed agent to run.

s79
00:06:34.000 --> 00:06:43.600
So let's say at 8:30 8:30 a.m. every morning before the market opens you want to go and you know do some kind of financial analysis before the market opens up.

s80
00:06:43.600 --> 00:06:45.360
You could set a trigger.

s81
00:06:45.360 --> 00:06:50.800
The managed agent would essentially wake up, run its task or run its input, output a result.

s82
00:06:50.800 --> 00:06:54.600
You can like do webhooks so that you could send a Slack message.

s83
00:06:54.600 --> 00:07:02.280
Maybe it queries your team's Jira board every morning and sends you know here's what everyone's working on and things like that.

s84
00:07:02.600 --> 00:07:04.560
So we talked a little bit about the harness.

s85
00:07:04.560 --> 00:07:10.280
It's the anti-gravity harness that runs under the hood and you can change the different models that it runs with.

s86
00:07:10.280 --> 00:07:14.680
Yesterday we just launched Gemini 4 Argon which is our new frontier model.

s87
00:07:14.680 --> 00:07:17.720
So I definitely recommend testing it out.

s88
00:07:17.720 --> 00:07:19.640
It's a great model.

s89
00:07:19.640 --> 00:07:23.120
But we'll show it later on in this in demo.

s90
00:07:23.160 --> 00:07:28.520
And you can set the different model in the config for the managed agent in the agent config as you see there.

s91
00:07:28.520 --> 00:07:31.280
We have that kind of one-line parameter to set the model.

s92
00:07:31.280 --> 00:07:35.000
So that's how you can kind of tune the managed agent based on the workflow.

s93
00:07:35.000 --> 00:07:39.480
If you're wanting to do something that needs to be fast and efficient, you can use Gemini Flash.

s94
00:07:39.480 --> 00:07:47.800
If you want to do do that's deep reasoning or potentially doing a lot of analysis, then you can use a different model like Gemini 4 Argon.

s95
00:07:49.600 --> 00:07:52.560
Like we said, Gemini 4 Argon was released yesterday.

s96
00:07:52.560 --> 00:07:53.720
Um it's a really great model.

s97
00:07:53.720 --> 00:07:58.320
It's especially good at financial tasks, which is great for hopefully for the audience here.

s98
00:07:58.320 --> 00:08:02.720
Um and really good also at doing data analysis across a wide range of documents.

s99
00:08:02.720 --> 00:08:07.400
Um and as always, the Gemini models are also super super good at multimodal.

s100
00:08:07.400 --> 00:08:14.160
Another really cool, interesting um feature that we have with the Gemini 4 model is long decode.

s101
00:08:14.160 --> 00:08:19.800
So, normally models have an output token max of around 64K tokens.

s102
00:08:19.800 --> 00:08:30.840
Um but with Gemini 4, we actually have up this so that now the 1 million kind of context window or token limit is now shared between input, thinking, and output.

s103
00:08:30.840 --> 00:08:40.039
So, output, as you need to kind of scale up if you need to do extreme kind of math reasoning or financial modeling where the output might be pages and pages and pages long,

s104
00:08:40.039 --> 00:08:41.840
the model can actually go and do that.

s105
00:08:41.840 --> 00:08:48.560
And this is showing to actually advance some of the more mathematical and financial benchmarks out there.

s106
00:08:50.960 --> 00:08:55.680
So, we talked a bit about the interactions API and environments, and this is kind of one of the key pillars.

s107
00:08:55.680 --> 00:09:05.680
These are kind of two core primitives in managed agents that allow you to get a lot of flexibility and be able to customize kind of the agent configurations that you do.

s108
00:09:05.680 --> 00:09:11.240
So, these are kind of two different um stateful things on our back end in our server side.

s109
00:09:11.240 --> 00:09:15.080
So, interactions is what manages your conversation.

s110
00:09:15.080 --> 00:09:22.720
So, when you put an input and get an output and kind of a different turn with the model, um those are saved as kind of what we call interactions.

s111
00:09:22.720 --> 00:09:29.240
So, you can take one interaction, pass that ID to the next interaction, and that kind of, you know, it's the second turn.

s112
00:09:29.240 --> 00:09:31.600
Um and then we also have environments.

s113
00:09:31.600 --> 00:09:37.440
So, environments are what's actually kind of in the file system on the sandbox.

s114
00:09:37.440 --> 00:09:50.480
So, when the agent goes and writes files or creates images or does different file tasks on the actual sandbox in the cloud, that's kind of representative of the environment.

s115
00:09:50.480 --> 00:09:56.680
And you can go ahead and use our APIs to download files from that environment or upload them.

s116
00:09:56.680 --> 00:10:06.240
But having these two things kind of split out allows you to do some really cool things where you can actually see here that we are creating one agent interaction

s117
00:10:06.240 --> 00:10:08.240
with our first code block.

s118
00:10:08.240 --> 00:10:11.080
And it's going to go ahead and do some work and save a file.

s119
00:10:11.080 --> 00:10:14.760
And then we can pass that environment ID to our next agent.

s120
00:10:14.760 --> 00:10:16.320
So, it actually picks up.

s121
00:10:16.320 --> 00:10:21.720
It has a very fresh kind of context window and conversation, but it's reusing the environment.

s122
00:10:21.720 --> 00:10:25.680
So, it can go and, you know, say save the Q3 filings in our workspace.

s123
00:10:25.680 --> 00:10:30.240
Maybe we're going to go, you know, fetch the fetch the annual reports, the filings.

s124
00:10:30.240 --> 00:10:33.560
And then we're going to go and do some analysis, save that as a file.

s125
00:10:33.560 --> 00:10:42.360
And then we can have a new agent come in, reuse that same exact environment, but this time have a fresh context window.

s126
00:10:42.640 --> 00:10:46.280
Kind of a multi-agent workflow, which is really cool.

s127
00:10:46.839 --> 00:10:53.680
So, there are a bunch of built-in tools already in the sandbox, like code execution, reading and writing files, as well as Google Search.

s128
00:10:53.680 --> 00:10:58.040
But we also support, you know, remote SMTP servers, so you can bring your own servers.

s129
00:10:58.040 --> 00:11:00.040
You can hook up to your favorite services.

s130
00:11:00.040 --> 00:11:08.960
And really that takes the agent a step further and be able to actually do kind of daily tasks or things that you can build into your apps.

s131
00:11:10.040 --> 00:11:14.440
And we talked a little bit about it, but you can also kind of mount files to the file system.

s132
00:11:14.440 --> 00:11:26.000
So, you can upload your different agents.md files, your skills, and kind of run through the playbooks that you have your agents doing on your personal machine, but doing it in an environment where you can kind of reproduce

s133
00:11:26.000 --> 00:11:29.560
and have that work done for you in the cloud.

s134
00:11:29.560 --> 00:11:34.040
So, you don't have to worry about, you know, overriding any files or messing up things on your machine.

s135
00:11:34.040 --> 00:11:40.080
Everything runs in the cloud, um and can just run as a process and get you back that output.

s136
00:11:40.600 --> 00:11:50.360
So, here's a Here's a code sample where we're actually going to upload an agents.md example into our um sandbox, and then we're also connecting a GitHub repository.

s137
00:11:50.360 --> 00:11:57.240
So, now you can actually have the agent look at your code, check for bugs and things like that.

s138
00:11:57.280 --> 00:11:58.600
And then you can also save the agents.

s139
00:11:58.600 --> 00:12:01.680
We also have an agents API, so you can do agents.create.

s140
00:12:01.680 --> 00:12:09.520
So, once you've kind of refined your workflow and found an agent that works for you, you can save it so you can easily call it, you know, in multiple different times.

s141
00:12:09.520 --> 00:12:13.560
You can iterate on it and then save it for later.

s142
00:12:14.160 --> 00:12:20.320
And then you can just easily invoke your personal agent by calling it with its uh agent ID.

s143
00:12:21.640 --> 00:12:28.640
So, another big part of the security aspect of managed agents is that secrets never actually enter the sandbox.

s144
00:12:28.640 --> 00:12:35.880
So, our managed agents have a proxy that sits uh it's a kind of a layer between the sandbox and our API.

s145
00:12:35.880 --> 00:12:46.480
And it actually what it does is it puts a placeholder kind of dummy token into the sandbox so that the agent can never kind of go rogue and say, "Oh, I see that I have this GitHub token laying around.

s146
00:12:46.480 --> 00:12:48.560
Let me just see what else is on there."

s147
00:12:48.560 --> 00:12:53.240
Um so, it can actually never do that because it never actually has your GitHub token.

s148
00:12:53.240 --> 00:12:58.960
So, what it does is when it goes and calls the GitHub API, our proxy will then inject that token.

s149
00:12:58.960 --> 00:13:02.839
It will match it to say, you know, it's API uh GitHub API.

s150
00:13:02.839 --> 00:13:06.160
Um we're going to go ahead and inject this secret.

s151
00:13:06.160 --> 00:13:11.800
Um so, that way the agent can never do anything that it shouldn't be able to.

s152
00:13:12.880 --> 00:13:22.120
We talked quickly about triggers, but triggers are also really powerful because you can have them set to kind of a daily briefing every morning like we said, scan your emails,

s153
00:13:22.120 --> 00:13:28.720
maybe suggest ones to reply to, or if you want to check on the stock market before it opens, um you can do that.

s154
00:13:28.720 --> 00:13:34.720
And this is a way to run one of your agents on a set schedule.

s155
00:13:37.160 --> 00:13:44.120
And we talked a little bit about multi-agent workflows, but managed agents can actually work really well um in multi-agent workflows.

s156
00:13:44.120 --> 00:13:55.640
So, you can actually set it up so that one managed agent knows how to, because it's just an API call, actually call other managed agents um and do work and fan out work.

s157
00:13:55.640 --> 00:14:01.480
And then you can have very kind of niche and specific targeted agents that have one core purpose.

s158
00:14:01.480 --> 00:14:03.520
They do that purpose really well.

s159
00:14:03.520 --> 00:14:07.360
Um you know, maybe it's drafting notes or doing research.

s160
00:14:07.360 --> 00:14:13.760
Um writes it to a file, gets that information saved, and the next agent can then take over from that different input.

s161
00:14:13.760 --> 00:14:23.640
And this kind of breaks it up so that you can have things really controllable and really make sure that you're doing specific tasks well instead of just giving a super super broad,

s162
00:14:23.640 --> 00:14:31.280
you know, task to an agent and having it have no way of kind of measuring what it's actually doing.

s163
00:14:32.440 --> 00:14:35.320
So, now we'll just quickly put this to the test.

s164
00:14:35.320 --> 00:14:40.040
Um so, we can actually run the agents here and see what this actually looks like when you run one.

s165
00:14:40.040 --> 00:14:44.680
So, we have an SEC kind of an SEC filing researcher that just researches companies.

s166
00:14:44.680 --> 00:14:47.000
We have it set to Nvidia here.

s167
00:14:47.000 --> 00:14:49.720
And we can go ahead and make call the API.

s168
00:14:49.720 --> 00:14:51.560
So, like again, this is just an API call.

s169
00:14:51.560 --> 00:15:00.839
It's really easy to build into your own either internal workflows or to actually publish in your services or apps to external folks.

s170
00:15:00.839 --> 00:15:03.600
So, we can go see here.

s171
00:15:03.600 --> 00:15:05.120
There are step traces.

s172
00:15:05.120 --> 00:15:08.960
So, each thing the agent does um is saved as a step trace.

s173
00:15:08.960 --> 00:15:21.120
So, you can see we're having function result for the right to file, code execution, and you can kind of trace the what the agent is doing and optimize as you go to kind of make sure that the workflow is straightforward.

s174
00:15:21.120 --> 00:15:24.480
And then were we using the same environment ID that we talked about?

s175
00:15:24.480 --> 00:15:28.680
So, because we're using the same environment ID in this scenario, we're actually having our agents pick up.

s176
00:15:28.680 --> 00:15:37.440
So, we can have We had this SEC filing researcher that went and did research on Nvidia, gave us a bunch of metrics as the output.

s177
00:15:37.440 --> 00:15:44.040
And then, we can now kick off this kind of quant agent, which will go take a look at the metrics because it's in the file system.

s178
00:15:44.040 --> 00:15:46.720
It's reusing the same environment, the same sandbox.

s179
00:15:46.720 --> 00:15:57.520
And now, it can pick up its task and its job from that um metrics.json, take it, run a bunch of Python code files to see and actually look at it make evaluation

s180
00:15:57.520 --> 00:15:58.880
for Nvidia.

s181
00:15:58.880 --> 00:16:08.280
And if we go here and look at the MD file, we can see the managed agent went and actually did research based on the metrics it got, and it's now kind of doing a forecast

s182
00:16:08.280 --> 00:16:11.880
for the share price in the stock.

s183
00:16:12.880 --> 00:16:19.600
And then, we can kick off a third agent just to kind of take that research and knowledge dump and turn it into a report.

s184
00:16:19.600 --> 00:16:32.720
And if we had connected different MCP servers, like we could connect Google Docs, have it actually output and write to a Google Doc, or send a message in Slack so that you can really kind of create these customized workflows from managed agents.

s185
00:16:34.400 --> 00:16:39.240
And then, you can also build managed agents super easily with the SDK into your own application.

s186
00:16:39.240 --> 00:16:45.800
So, this is just a demo here where we have a managed agent that is actually um connected to our service.

s187
00:16:45.800 --> 00:16:56.120
Um it's connected to our Google Drive where we dumped in a bunch of receipts of our annual kind of um company data, and it's going to go ahead and actually create this visualization

s188
00:16:56.120 --> 00:16:58.680
um by from from the the data.

s189
00:16:58.680 --> 00:17:01.560
And the managed agent is the one that's actually doing all the research.

s190
00:17:01.560 --> 00:17:07.839
It has, you know, kind of a skill that will go out and tells it kind of how to do this interactive report.

s191
00:17:07.839 --> 00:17:13.439
Um and then, it's just kind of reasoning through it and building this out for us.

s192
00:17:17.280 --> 00:17:18.319
So, it's really customizable.

s193
00:17:18.319 --> 00:17:33.040
You can really kind of anything you can do on your own personal coding agent, you can kind of turn it easily into a managed agent that you can either run as a company daily or set up in your own apps so that we have other ones on our on our Gemini API on AI Studio where you can

s194
00:17:33.040 --> 00:17:35.560
kind of create games on the fly.

s195
00:17:35.560 --> 00:17:39.920
You can research documents and do all kinds of things.

s196
00:17:44.880 --> 00:17:50.480
So, the easiest way to get started today is to go over to AI Studio and get started for free over there.

s197
00:17:50.480 --> 00:17:57.640
And we also have our documentation for managed agents, but I'll quickly just do another demo here since we have a few minutes.

s198
00:17:57.640 --> 00:18:03.800
This is one of our kind of managed agents examples that is on AI Studio in our gallery.

s199
00:18:03.800 --> 00:18:10.320
And this one's kind of just trying to showcase how you can actually build these managed agents into your own applications.

s200
00:18:10.320 --> 00:18:15.200
So, this is you know, create a nice front end and put in a stock.

s201
00:18:15.200 --> 00:18:17.280
Oh, not slash.

s202
00:18:17.280 --> 00:18:19.760
So, we can go ahead and do VOO.

s203
00:18:19.760 --> 00:18:29.480
And this ticker agent is a bunch of our different managed agents that will go out and do specific tasks and look for different analysis on a stock.

s204
00:18:37.280 --> 00:18:45.800
And again, this is kind of more of an illustration illustration that's like as you're building managed agents into your own apps, you might not want to have this kind of stream

s205
00:18:45.800 --> 00:18:49.160
but just to kind of visualize what the actual agent is doing.

s206
00:18:49.160 --> 00:18:53.080
It's reasoning through all the different steps in the interactions API.

s207
00:18:53.080 --> 00:19:00.080
And then this managed agent is told to kind of generate a report based on all of the findings and reasoning it did in its sandbox.

s208
00:19:00.080 --> 00:19:10.040
So, now if we go ahead and click on the report, we can see the managed agent outputted this kind of PDF file for us where we can see all of the research it did,

s209
00:19:10.040 --> 00:19:15.040
see its analysis and key takeaways, as well as doing these charts so we can visualize.

s210
00:19:15.040 --> 00:19:24.360
And it's also going to go and link out to kind of all of the resources um that it went ends actually using for this.

s211
00:19:24.360 --> 00:19:32.360
So it actually went to the Vanguard Vanguard site, found the different documents, read them in, and then did analysis on it.

s212
00:19:33.880 --> 00:19:36.440
And that is it on managed agents.

s213
00:19:36.440 --> 00:19:38.840
Thank you very much.