Hey, this is Jacob from LinkChain. So,

we've just released Linksmith Agent

Builder in public beta. Agent Builder

allows you to build production grade

agents without writing any code, just

using chat. And over the last month or

so, since we first released the product

in private preview, we've seen people

build a ton of different agents. And one

of the really common use cases that

we've seen is building a research agent.

You know, many of us in our daily work

need to do some kind of research,

whether it's, you know, as a marketer,

I'm doing market research or competitor

research, product research, or if you're

an AE, you might be doing customer

research.

And so, I'm going to show in just a few

minutes how we can create a market

research assistant to help us with um

some of these tasks. So, the first thing

we need to do is just give agent builder

a prompt.

So now agent builder is researching the

tools that it has available and the

triggers and determining what it needs

to accomplish the task.

Now the next thing is uh we need to

follow up with some clarifying questions

because as you might have noticed my

initial prompt was pretty basic. Um so

in order to create the kind of detailed

prompt that we need for an agent agent

builder is asking us for some additional

details. So, what specific aspects are

we interested in? Let's say um technical

trends.

And how about companies? That sounds

good.

How often should it run? Let's say on a

schedule.

How would I like to receive it? Let's

say with a Slack

direct message.

Um, how about company

announcements

and Reddit?

Let's get this as a weekly report.

And I'll put my Slack email address in

here.

Cool. So now agent builder can go ahead

and configure our agent. All

right. Now, we're just about ready. And

we can create this agent in a shared

workspace for our whole team. Or since

we're just uh going to be testing this

one out, let's just make this available

for me. And it's sending at 6 a.m. on

Monday. That's good.

All right. So, now we're in our edit

mode. And on the left hand side here, we

can see the detailed prompt the agent

builder has created. So it has a role,

its responsibilities, its key research

areas and how it should actually go

about conducting the research. On the

right hand side here we can see a

visualization of our agent. So we have

the trigger, our supervisor agent and

then two sub agents um that's are

specialized in doing research into

company and fundraising uh research and

technical trends. You can see both of

them are armed with a Tilly web search

uh tool and we can down here we can see

our trigger and the toolbox available.

Now here we got our test chat and in the

upper right hand corner here we've also

got this require tool approval um

option. And so this is in our debug mode

we can go ahead and require tool

approval before it actually goes off and

calls it new tools which is useful as

we're testing things out. Um now we'll

try running this agent in just a moment

but um sometimes these research tasks

can get pretty big. So, in order to keep

this um right sized for the kind of

insights I'd like to get, which is just

sort of the top three news stories of

the day, um let's go ahead and uh

provide some feedback to our agent to

update our system prompt. So, how about

only give me the top three items you

find each week.

Okay. So, now agent builder can go

through and update its system memory.

We'll approve that tool call.

Okay, great. So, there are a few places

that um needed to be updated in the

system prompt. Um we got a confirmation

message here. Now, we're ready to go.

So, um this will run on a weekly basis

next Monday, but um let's go ahead and

see what we get when we run this now.

All right. Now, Agent Builder has

completed the research and it's ready to

send a Slack message. I'm going to click

continue in a second, but I just want to

show you this took a little bit longer

than what we show here in the demo.

That's because it's going off and doing

tons and tons of research from lots of

different sites, compiling all this

information, all to synthesize this down

that is concise down to the top three

items. So, we'll go ahead and click

continue here. All right. So, now it's

been sent. All right. Right. So now we

can see our weekly update here in Slack.

We've got our message from agent builder

for the week of November 25th to

December 3rd and the top three

newsworthy developments, one on AWS,

Anthropic, and Run Layer with quick

synopsis and a link to each item. So

that's how we can create a market

research agent in just a few minutes

using Langmith agent builder. And we've

got this all set up to run in the

background every week and send us weekly

report on the top three newsworthy items

in Slack where we spend a lot of our

time. You can try this out for free

yourself by heading over to

langsmith.com.

Thanks for watching.