👉 Try Viktor with $100 in free credits (no credit card): https://ref.viktor.com/jlaw-yt Use code JONLAW100. This video is sponsored by Viktor.
Viktor is an AI employee that lives in your Slack or Microsoft Teams. You tag him like a coworker, and he does the work. He connects to over 3,200 tools, including Gmail, Google Drive, Notion, and HubSpot. He also has his own cloud computer, so he remembers context and gets better over time. In this video, I explain the shift from chatbots to AI agents that own real work. Then I set up Viktor in about two minutes and show how to permission integrations safely. Finally, I run four demos on my real business: a sponsor deal audit, inbox triage with drafted replies, team channel workflows, and a custom pipeline dashboard. Drop a comment with the first job you would hand to an AI employee.
CHAPTERS: 0:00 Viktor: The AI Employee That Lives in Slack 0:55 Video Roadmap 1:18 Chatbots vs AI Agents: Assigning Work, Not Asking Questions 2:08 AI Agent Permissions at the Enterprise Level 2:32 What is Viktor by Zeta Labs? (Slack & Microsoft Teams) 3:02 Viktor’s Cloud Computer, Memory & Learning 3:18 3,200+ App Integrations (Gmail, Google Drive, HubSpot) 3:37 One Shared AI Employee Working in Parallel Across Teams 4:06 How to Set Up Viktor in Slack 4:47 How to Get $100 of Free Viktor Credits 5:24 Connecting Google Drive & Notion From Slack 6:05 Connecting Gmail in the Integrations Dashboard 6:43 Viktor Dashboard Tour: Scheduled Tasks, Skills, Spaces & Docs 7:24 Talking to Viktor Through iMessage 7:42 Setting AI Agent Permissions (Auto, Ask First, Disabled) 8:35 Running a Permissions Audit on Your AI Employee 9:20 Demo 1: Finding Blocked Sponsor Deals Across Slack & Gmail 10:20 Demo 2: AI Inbox Triage, Company Research & Drafted Replies 11:12 Reviewing & Approving AI Email Replies Before Sending 11:37 Demo 1 Results: 23 Overdue Deals Indexed 12:15 Demo 3: Team Channels, Editor Briefs Doc & Slack Reminders 13:15 Finance Use Case: Invoice Tracking With QuickBooks & Stripe 14:00 Demo 4: Building a Sponsor Pipeline Dashboard Web App 15:27 Live Pipeline Dashboard & Weekly Scheduled Summary 16:21 Tip 1: Token Cost Scales With Task Difficulty 16:39 Tip 2: Onboard Your AI Employee Like a New Hire 17:23 Tip 3: Review Work & Expand Permissions Over Time 17:31 Tip 4: Viktor is a Generalist AI Agent 17:47 Final Takeaway: AI Employees vs Chatbots
------------------- Clips channel @jonlawclips
I do NOT sell any products of my own. Watch out for scams.
Victor is an AI employee. He lives in Slack.
He's connected to over 3,200 tools. And when you hand
him a job and work, he just does it.
And in this video, I'm going to explain what an AI employee
actually is, how to set one up.
And then finally, we're going to run through four examples of real
work that an AI employee can do in our business.
And then by the end of this video, you're going
to be able to replicate those kinds of results
in your own business in the pursuit of becoming
I mean, AI native such that your product
productivity, your speed of iteration
and of innovation is enhanced through these tools.
We live in a pretty crazy time where it
is actually possible to get a lot of your work
just offloaded like this with these agents.
And so I'm excited for this video. It's going to be
a fun one. I hope you enjoy it as well.
This video is sponsored by Victor and everything
you're about to see is me running him on our real work.
And that said, let's get
into it so quick roadmap first i want to level set
with what the system is that we're going to
use then i want to walk through the setup it's about
a two minute process that you can do as
well to set up one of these employees inside your business
and then we're going to walk through
four demos of again real work that you can do
with these agents and also explore some
of the different features and how to get
The most use out of the system.
And now I think there is a little bit
of a modality shift that has to occur
where the kind of standard pre-agent way,
which so 2025, you know, right?
That frame of mind for engaging with AI is the chatbot modality.
It's asking.
a question to a system that has data it's saying like what
was the average deal size last quarter
or tell me four important things that i need
to do this week based on this set of information
now what's different about that frame
and where we are now with agents is agents are
who you assign the work to as opposed to just ask the questions too
so as opposed to saying
hey what's the statistic just say hey go build
the report give me a deck give me a
Report that I'm going to give to my team at our weekly all hand.
So that's a shift that takes some time to kind of mentally grow into.
And there also is a very
Real permissions side to this at the enterprise or the corporate level.
Obviously, you don't want to just give AI everything immediately
and just send it to do all of your work.
You want to grow into it over time.
And that's something I'm going to continue
to emphasize throughout this video
and illustrate how you can permission correctly your integrations,
your data and the access of the agents.
So that said, who is the Victor I referred to?
Victor is an AI agent, an employee made by Zeta Labs,
and he's an employee.
that lives in either Slack or Microsoft Teams.
So wherever your business already communicates
and does work.
And you engage with Victor
by just kind of at mentioning Victor,
just like any other employee or someone on your team.
So you go,
Hey, Victor, go do this for me.
Hey, Victor, go pull this.
Hey, Victor, go send an email on my behalf, whatever it is.
And then Victor, this employee has
his own computer in the cloud where he learns over
time, he'll store data, he'll recall conversations,
and he will get better at the tasks he was
assigned because his context is kind of.
Now, like I said,
the AI employee is connected to about 3,200 tools.
So if you go engage in HubSpot and Gmail and Drive,
all those different tools or apps,
the agent can as well natively and can interact with them.
And that's really how you get the extension
of the intelligence into doing real work,
because it can just like you would.
And then Victor as well is shared.
What that means is.
Everyone can be adding
or sending tasks to Victor at the same time, and the agent can
be going and doing all of that work simultaneously.
So it can be in the finance chat and the PR chat
and your kind of executive chat, whatever
it is at the same time.
Doing those work for you on your behalf in the cloud in parallel.
So that said, let's jump into my computer to do the setup
and then we're going to get into the work walkthroughs after that.
So this right here is the Victor homepage.
You can do this just directly through Slack
as well through their app store.
But I want to show you what the setup looks like here.
I think it visually is cleaner.
So let's click continue as John at this point.
Just go ahead and create your own account if you're following along.
Once you press that, you're going to get to
This Slack screen,
I'm going to go ahead and enter john-law.slack.com, we'll continue.
And then here's where we're going to let Victor.
access our single channel and then we're done.
The integration is pretty much that easy. Okay,
so there you go. That is the setup
and how to get the whole system installed. Now we're going to
jump into those walkthroughs or actual work that the agent is doing.
You're going to learn better
from this video if you go ahead and set it up and do these tasks,
these demo tasks at the same time
as I do. So if you want to have your employees set up
by the end of this video in your business
And you support the channel while doing so.
And the link is in the description.
to do that, you can use code JOHNLAW100
and you will get $100 of his work free running in minutes.
So again, you get $100 for free,
you support the channel, and you get to try out.
Victor, with all that said, let's jump into my computer
to get started with the and now if we jump over here into Slack,
we can see our direct message channel was set up with the Victor agent.
So what you're going to immediately want to do at this point
is make sure that the agent has the context
and the connections that needs to do work on your behalf.
So if you if your work lives in Google Drive, if it lives in HubSpot,
wherever it is across the internet, that's what you want to
Connect next and we can do that in one of two ways.
Both of them work.
The first is just by talking to the agent natively inside of Slack.
I'm going to message Victor.
I'm going to say connect our Google Drive
and notion and tell me what you can see.
We can see Victor has started working and while it's working on this.
I want to jump over here into the dashboard and the second way
in which we can connect the app is by jumping into integrations.
And then here I can see email and Notion like, for example, on Gmail.
Let's click everyone and continue to Gmail.
And then here, for example, I can connect my Gmail account.
Now, if we jump back over here into Slack for that option, one,
I can also click, for example, connect Notion.
It'll bring me over to the integrations page. I'll click.
Connect Notion, continue to Notion.
And let me go through that full flow one more time
and actually connect an account.
Okay, we successfully connected Gmail.
And now here in the app dashboard,
this is where we can see all of the integrations
like we've already done for Victor.
We can see scheduled tasks
that you want your agent to go work on
while you're not interacting with it.
The skills, so what Victor can do based
on work that it has done for you before.
So basically learning over time
and getting better at the work you assign to it.
We can also see our spaces.
This is the development side of Victor.
So building kind of full stack apps that can be public or private.
I think public is more useful for you.
And then finally, the docs.
So this is kind of workspace collaboration that is AI native,
something that's really fun as well that we haven't touched too much.
On this channel is iMessage and this is literally being able
in addition to Slack to and this is being able to communicate just
with your iPhone with Victor telling it to go do stuff on your behalf
in addition to Slack, but we are going to focus on just Slack.
for this video.
Now permissions is going to be very important
for all of this, right?
That's a big point for all of us existing
on the more enterprise side of integrating AI
into our workflows.
Now, like you saw when we went and connected Gmail,
you can keep that just private.
So just between you and the agent
or make it available to everyone
who can interact with the agent.
You wanna be careful with that permission.
And if you ever wanna change those permissions,
you can do that also by just talking to the agent.
So I can say,
Hey, Victor, go change Gmail,
send email permission to ask first as opposed to turned on by.
OK, and in this case, it looks like it was already set to ask first.
So that was an easy one.
Let's say what are other options for this permission?
And we can see here it looks like we can have it
as auto ask first or disabled.
So I think ask first works there.
And what we can also do at any point is say something like run
an audit of permissions you have and who else in my team can access it
So I think something that this system does very well specifically
is regulate the kind of access that you and your team members have
so you get all the benefits of an AI native collaborative ecosystem,
but also don't get some of the downside
of any privacy concerns that you may have.
And so what I'd recommend is just start out by default on something
like ask first and private and then over time as you get comfortable.
And also based on the level of security you want,
like say Gmail versus all or
versus software tool that the team uses.
Over time, give more responsibility to the agent
as it learns how to do things the right way
by actually interacting with you and your team members.
And so now for the first demo here,
I wanna give Victor some actual work
and let's see how well it does there.
And so what I'm gonna say is this.
So go through this channel and the shared Gmail,
find every sponsor deal from the past two months.
That's still waiting on something from us
and then list the blocker and who
owns it. This is a really big task, right?
A lot of deals in that pipeline.
So I want to see, okay,
how well can it find all of that number one
and then complete that work of
pressing it and making it understandable, identifying the blocker.
And giving that list to us.
This is something that would take an intern,
you know, five hours, a full plus, a full
workday plus to complete.
And let's see how well Victor can do.
Now, what I want to do now, while this is working,
I want to jump back over here into
integrations.
What I want to connect is Google Drive.
So let's go ahead and integrate this.
Okay, now Google Drive is connected.
Now what I'm going to do simultaneously
while the agent is working on summarizing to-dos over here,
is jump into the web app. I'm going to click on new project.
I'm going to call it email. And then
now we have Google Drive and Gmail connected.
So I'm going to say this. Go through the partnerships
inbox for the past seven days.
And then for the real ones, look up the company, know what they
sell, and then draft a first reply in my voice using
the rate card in Drive. And then put every
draft in this thread to approve. Don't send anything yet.
Over time, you can have the agent
And again, just go do it on its own.
But for now, we want to be...
A little more careful here we can see it's
checking existing sources and preparing that
over scratch that remove that.
Okay, so we can see that it found about 24 threads. Okay,
we can see this first email project has wrapped up.
So we have about 24 threads in progress,
then it's gone ahead and drafted the responses to those three,
we can see the email there.
And then I think what I really want
to double click on right here is the
UI UX layer. So it didn't just draft the email that I have to copy
and paste into Gmail,
but it made it in such a way that I can click send reply.
And then the agent will go ahead
and do the work or send the reply on my behalf. But I still get.
Importantly, right, to review that.
So again, we get that permission layer
and we still get the benefit of AI doing the work on our behalf
Okay, and we can see the same thing.
Okay, we can see that the last two months of deals
have been indexed in here.
So we have 23 deals that are due
Okay, so we can see that now back in Slack
Victor has also indexed the currently due or overdue videos.
It looks like there's 23 that
are due or overdue, which is not a great reflection
on me as a YouTuber. And it looks like there's over
50 other deals in the pipeline here that are waiting on contracts
or invoices. But again,
that's something that's six or seven hours of work was done here by
thank the heavens, Victor.
But now this is just my use case say if I'm
the creator kind of the manager in this context but
let's say we have a channel I'll create
a new channel I'll call it editing and let's say we
use this to discuss different projects
that are being edited with the actual editors. If I was an
editor in here I'd just jump into this channel
in Slack where we already talk about this I would
at Victor and then say something like
this go through recent briefs in the drive folder since
we connected Google Drive build a doc with the videos
that i will need to edit as the editor
and then remind this channel and Slack
so hashtag editing two days before each one is due so this is
something so it's both doing the work for us now
and pulling out the information we need and then
also in the future it'll just go and ping the Slack channel
and here's where you can start to
see how this will do work for you, but also become much more of the AI
kind of native collaboration backbone for your business
and your team over time.
So let's send that message and we'll invite
and we can see Victor is working.
And for your finance team, for example, you can do the same thing.
Let's say you have a channel called Finance.
What I can go ahead and do if I was the CPA for this business is say
Hey, pull every invoice we sent in the past 90 days
And that's if you've connected QuickBooks or Stripe
Match them against payment received, draft reminder emails
So once more, the underlying theme here
is an AI native collaborative agent that is both
permissioned, so safe for your business,
but can also give you all the juice there is to
squeeze out of agentic
Automation and intelligence.
Okay, and if I click on this,
I can see a actual Google Doc that was created
that has all of the videos
that me as the editor wants to edit.
Okay, now for the last demo here,
I wanna go above and beyond,
tie everything we've been asking Victor to do together
into building a app, full-on add-on system
that's custom for us to do and better track.
The work that we've been asking it to do prior.
Now we can do this over here in the web app.
We can also just do it still natively through Slack.
Let's go ahead and see what that would look like.
I'll jump back into my DM with Victor and I'm going to give it
this kind of lengthy prompt,
which is build a sponsor pipeline dashboard
as a web app, pull the deals from our prior conversation.
And then this is important as well.
Set up a recurring job such that once per week you post a summary
of that pipeline or that app and the three most urgent follow-ups
and let's call it in the social
channel and then ask me again permissions before
you connect anything new now we can go ahead and
do that from Slack just to illustrate the build let's go
and build this in the actual web app
let's call it operations tracker
and i've given it this prompt and keep in mind that this can
kind of bounce back and post things in Slack as well
so this the Slack and the web app
Basically are one all that matters is how you want to talk to the agent
and it will go ahead and do the work across the whole ecosystem.
We can see it's building the dashboard
with the 23 deals that it found in.
the gmail okay we can see it's now writing the
code okay we can see the pipeline dashboard is
live with 300 grand in deals press on that so we have our live web app
showing these deals by stage looks pretty nice
and actually i am a fan of the UI UX here and we can add a deal like so
now if you want to update this or share it we can
do that with the link and by
just talking to Victor and everyone
in the team can engage and then we've all
And then now if we jump over into scheduled,
we can see the task that was set up.
So every week it's going to go ahead and run that.
And so I know we've gone through a lot here,
but I want to wrap this up by going back to
Victor and just saying in the spirit of friendship
and future AI overlords, Victor, you're the
best.
And that is the bottom line here that I want
to illustrate of the potential when you and
your business and your team integrate natively with AI.
Okay, so there you go. A few things I want you to be aware of here.
Number one is cost or basically
token usage will scale with the difficulty of the task.
So if you throw a bunch of really hard tasks
all the time at the agent, that's going to take
a lot more tokens versus more mundane kind of tasks
that are standard, say white collar work.
Now, second, the week one kind of the onboarding is
on you and giving the agent everything it needs to do well, right?
Treat it like an employee,
Give it lessons, give it context,
show it how to do things. If you do that,
That's how you really natively become AI powered and supported.
It's not just by throwing the agent in there
and just letting it kind of do its own thing.
You need to give the context and the information
and the training to the agent because it remembers
that and that's the basis for everything moving forward.
So you can make that a much cleaner
and quicker process and get higher quality work out of
the agent.
By doing that just a one time thing in that first kind of week
when you have the employee,
just like you would for a real human employee.
Now, third, I would still review things this ties back to permissions
and then just shift over time into giving the AI more freedom.
And then fourth and final point, Victor is a generalist.
If you want a specific solution because your business
does one very specific thing just over and over again,
that's going to be a little bit of a little bit different.
of a setup, and that's something that we have covered
in depth in other videos on the channel.
So if I could impart one thing in this video,
it is that fundamental difference between
chatbots that just answer questions and AI employees,
where I think the industry
and businesses are at,
own actual work streams that are repeatable over time
and based on context,
a context pool and awareness that just grows over time.
and makes these agents more powerful on your behalf.
Victor is a pretty cool system and kudos to the team that made them.
I hope you get a chance to play around with the agent
and integrate it in your business
and see what these kind of systems can offer.
That in mind, appreciate you being here.
I will see you in the next one.