#15: Preparing Your Workforce for AI
Welcome to another episode of The Junction.
I am so excited to talk about AI committees today.
I think it's really important that people
understand that we are not just running
these things in a vacuum by ourselves.
You are not alone.
And if you are wanting to implement an AI
strategy at your organization in that a committee going
about it by way of committee could actually be
a really good strategy for you.
So we'll talk about that.
We'll jump into our headlines, per usual and
see where the conversation goes in between. Sound good?
Just make sure we cover all the bases.
But we don't go too far because sometimes we go
too far and sometimes we don't go deep enough.
So let's meet.
Are you getting feedback from our listeners on Know?
There was that one guy, though,
that reached out and said hello. Shout out Jason.
Yeah, Jason.
Super cool for him to reach out, say hello.
I think admittedly he was our first fan mail.
I don't know if we said that we
were pretty excited about our first fan mail.
We were super pumped and it was very thoughtful.
Jason, I just have to shout
out like very thoughtful email.
I think I actually would like to in our next episode,
sort of like there may be some copyright infringement on this,
but I want to play the song that he sent over
YouTube and kind of break down his take on the land
of stable diffusion and all the things AI.
Yeah, sounds great.
So today we're going to talk about we have rolled
out established an AI committee here at Venn Technology.
How did that come to know?
I think, like some of our podcast
episodes, it just came on a whim.
And I realized that I think I
was the one that started it.
And if I'm not, somebody can tell me. I think you were.
Yeah, but I realized that in what we're doing,
we're doing a lot of top down talking like,
hey guys, we're going to go do this.
You don't have a lot of opportunity to buy
in or come up with it on your own.
And in all of these discussions we're having, it's
really important for everybody to be bought in. Right.
Because we're talking about automating people's jobs.
We're talking about how do we be more productive?
And you have to have the sense of and the whole
organization needs to buy into what we're doing because from the
types of automations that we're building out, it's not just like,
well, let me help me write my resume.
We're talking about across all of
the processes that we have.
Like AI is going to touch all of them and how
can we do it in a way that helps you, but
also not only makes you more efficient but more productive?
Right.
And not in the, oh, well, the company is going
to try to extract more value out of me. Right.
But some of our goals are rewarding work.
People want to be excited about it, and people want
to put their hands on it and talk about it.
And some of them can't necessarily program,
but they have some really good ideas.
So that's where it kind of came into mind
because I've been talking to a lot of interviewees.
One of the questions that I've been asking
is, hey, where is this OpenAI thing going?
Maybe you haven't touched it yet. Maybe you have.
And they all have some really great answers because they're
trying to be their best self for the interview, and
they're all really creative, and I'm like, wow, I wasn't
thinking about oh, man, I didn't think about that.
But getting everybody's take on it has been
probably the most beneficial thing because everybody has
an idea on how it should work, and
nobody has been doing this for decades.
I mean, okay, maybe there's like,
one guy, but not on average.
Yeah, I like how we've so,
first of all, it's voluntary, right?
So you led a lunch and learn on this and
talked about here's kind of what we know, State of
the Union, where we think the opportunities are for use
cases, but then opened it up to the team.
Hey, if you're interested in being a part of
the AI committee here at Ben, raise your hand.
And then we established a cadence of meetings.
We have kind of your typical we've got
a few sponsors, kind of executive sponsors.
We've got folks that are making sure we're
taking down the notes, driving the meeting agenda.
But to your point, it's not just folks in
there that know how to code like myself.
I can talk all day long about use cases, but
once you all start jumping into the tool and we
even did some whiteboarding the other day, which I do
like that I love me a whiteboard doodle.
But when we start jumping really deep into the weeds
on the architecture and how the solution is going to
work, you've kind of lost me a little bit.
But it's important that we have
that rounded kind of team, right?
So it's not just folks and from cross
functionally different departments talking about weighing the pros
and cons of if we pursue this use
case, what is the impact cross functionally?
Is it only benefiting one team
member or ten team members?
Is there a play to go to
market with this solution at some point?
So we're actively discussing all these things and then holding
one another kind of accountable week to week on what
we say we want to go do or build and
actually getting down to the action items of implementing it.
Because I think if it was just you, Chase, like, okay,
we're going to do an I mean, you happily go and
build these things on your own, in your own free time,
but you've got so many other hats you're wearing.
Right?
And so that's part.
One is just evening out the workload, but then two,
getting the buy in from the rest of the team.
And then three, that's something that people who are naturally
inclined or passionate about, they can have a voice and
it's kind of like a fun little side project.
Yeah, well, what's really fun about it is that
because for people have been in for a long
time, this isn't like brand new, right.
But I'll call it for the last six months, right.
People were there's an it's been more
readily available to your average consumer. Yeah.
Like, I talked to somebody the other day, he's like,
wrote a thesis on this like ten years ago. Yeah, okay.
He's probably sitting back like, Gosh, Mel.
All these people talking about if Mel would just
call my name out, people would know about me.
We are going to get him on the podcast.
Oh, that'll be cool.
Oh, yeah, we talked about that.
But I think what is really interesting for the
average individual, you have this wonder and excitement about
something that really, for the most part, the majority
of the population in the world has never really
talked about, done anything with. Right.
You think about being a mechanic and talking about
new mechanical processes and I'm already falling asleep.
But I do love my mechanic and I love working on cars.
But this has been around for a century, right.
Large language models and this AI the things that
have happened the last six months are like everybody's
starting basically from the same page, right.
And there's not just this one expert that's going to
walk in and be like, oh yeah, check me out.
I know everything, Mel.
I know more than like, everybody can walk in
and if they're up to speed on the news,
we're all operating from the same base.
So I think that's one of
the things that's really exciting.
Part of the benefit of having this committee is
like, we can all throw all these ideas out
there and some of them might land flat.
But some of them, I feel like, are going to be
not only really beneficial for us, but things that we can
package up and we can sell to our clients or educate
them on or consult them on how they're doing their things.
And I thought that was just super unique, that people
can walk into that room and not have to be
a coder, but can have some really great ideas.
And it was around that kind of continuous learning, one
of our core values and rewarding work, that I think
that the committee is really going to shine.
Yeah, absolutely.
So if you're thinking about starting a committee,
what are a few recommendations that you would
have around forming it cadence, things like that?
I realized quickly that at least as you
and I were going to be joining, we
weren't going to be joining a whole lot.
I think I missed two or three
of the meetings because of travel.
You definitely need to have somebody that is not only
excited about it, but can definitely keep the committee going
and has some level of organizational buy in.
Shout out to Troy. Troy.
Troy on our team has been instrumental in
keeping the wheels turning on our committee.
As Chase mentioned, we've kind of been in and
out for travel, know the occasional conflict, especially because
Chase, you still manage a pretty customer facing role.
So you know those client calls,
you got to sell those deals.
Yeah, we're going to sell them.
What the committee is coming up with. Yeah, exactly.
So I think cross functional, yes.
It should not be all leadership.
You should definitely have somebody who somebody one
or more people that sit in leadership so
they can communicate back to the leadership team
in the terms in which makes sense.
We have a regular cadence of meeting as a
leadership team and making sure that that basically is
communicated back what's happening, what's the impact, and then
having folks that sit in some of the more
tactical seats they're technically inclined.
And then you've got the folks that maybe are
basically you want someone who's touching pretty much every
kind of point of your customer lifecycle.
Probably you might just be focusing internally on internal
use cases, but there's a lot of value that
comes from having folks from basically every function or
department, which I don't actually know.
Do we have somebody from
every single function or department?
I think operations is not in on it yet. Okay.
But the other two things that actually came to mind were
having sort of a kind of a legal game plan.
Like first and foremost, lay the
ground rules for everybody, right?
What kind of data can we throw in here?
What are our agreements with our customers?
What are our own internal policies
on stuff that we're generating?
Can I just drop some code
that I wrote that's production code?
Can I just throw that in there? Is that cool?
Mel, are we good?
So come up with some game rules before you
start the committee and then after that I would
also focus on and if you have a legal
department, have someone sit in on it.
Oh yeah, that'd be great.
Sometimes they can, I guess, depending on and I don't
want to paint with broad strokes, but as long as
everybody comes to the meeting with the mindset that we
are brainstorming and this is an open, collaborative space and
we're not going to implement anything we've talked about today,
because that'll get your legal team fired up real fast,
forward looking statements, all those.
I also feel like we just earned some kind
of some mail the junction@ventechnology.com on this episode.
You just talked about how and as a lawyer
and I really like lawyers, I love law.
Did really well in my law class in business school.
Anyway.
Love you guys the other thing, too. They're important.
They protect, right?
Yeah, they aim to protect. That's awesome. The brand.
See, now we're going to get some good Fed mail.
That's always my spin as your
self appointed public relations coordinator.
Yeah, I love it. That's great.
Okay, so I felt you were going
to add something else to that. Yeah.
The last one is if you're in the committee
or you're on the committee, don't recommend or pull
technology or platforms that you don't have access to.
Everything that we talked about
or we are talking about.
We already have an OpenAI account.
We've already got salesforce, we've got zoom. Right.
We're building ideas around tools
that we're already using.
The challenge with doing it on something else,
like, I don't know, if you don't have
salesforce, and it would be like, great.
What if we did all these great things and we
implemented salesforce and we did all this other stuff?
Well, you just sign yourself up for a huge chunk
of work, and you don't even know it yet.
So focus on tools that you're already
spend and a lot of spend. Yeah.
So maybe not keep that as a wish list.
That could still be something that you
kind of have a backlog of.
But I do kind of like the idea of and
we've not incorporated this into our AI committee cadence, but
what if we spent five or ten minutes every time
we met on introducing, like, a show and tell of
one new tool, like a new AI tool?
I mean, these tools are rolling
out, like, hundreds, thousands a day. I don't know.
Ridiculous.
So maybe we do something like that.
We incorporate a show and tell totally component, because
even if it's not something that we use in
our use cases or solutions, it's kind of opening
a dialogue up around what these other tools are
doing, and it might spark some inspiration in how
we can apply some similar logic to our solutions.
Yeah, that is a great leeway.
Right into this, I'm going to shout
out this one company that I found.
They've actually been around for a
while, but it's called Fireflies AI.
And what was really unique is that it's
a call recording type set up where it
records the call, does the transcript deal.
It summarizes some stuff, it pulls out some bullets.
They've even implemented this kind of chat GPT.
They call it Fred GPT, where you
can ask questions of the transcript.
But I thought what was really unique is that they
basically invite you as the like, if I have an
account and you're on my meeting, I can have it
send you a summary of the call.
And it's like this automatic invite deal.
It's a genius business case for these guys to
just like, hey, come check out the call summary.
Oh, by the way, if you really like
this, you can sign up right now.
Just click this button.
Yeah, some good product marketers. Oh, totally. Oh, man.
I was like, oh, man, how much is it?
Now I got to figure out how to buy it.
But yeah, tools are coming out all
the time and I love that idea.
It's just constant.
Like we'll check this tool out, check that tool out.
It's not too dissimilar.
I mean, it's not the same, but you know, like
calendly anyone can book you don't need, you know, you
get people so conditioned to that, clients conditioned to it.
Now everyone's like, send me your calendar link.
It's like the new Kleenex. Yeah.
Yes, totally.
It's like the verb.
So what is the differentiator so fireflies,
they essentially integrate to your meeting platform.
Like zoom. Yeah, zoom.
They've got integrations into salesforce.
They've got a number of them, actually.
But I like your point though, right?
On your committee, or even if you're not going to run
a committee, it's more important to figure out what the tool
does and the thought or the idea behind it than it
is to sign up for a free trial.
Because this one deal that I pulled up,
like the idea of what automatically kind of
invited me by sending out a summary email.
Maybe you're just doing this
for your internal stuff, right?
Like send them a summary of the
transcript just to the internal folks. Right.
Was that idea that quick idea of getting buy
in from other people by utilizing this automation?
Anyway, I like how just an initial review
of their homepage It's groups you can filter
by task dates and times, questions, pricing.
That's really neat. Yeah.
Anyway, seems like a great group of folks, but I
feel like all of these if you pull up their
use cases, these are all ideas that we've been kind
of talking about just in one form or fashion.
So I wouldn't be shocked if they
have their own AI committee, probably.
And that's another thing that we are constantly evaluating when
we're looking at these different use cases is do we
build it or do we buy it or do we
wait for them to roll it there's?
We've talked a lot about the transcripts and the
sales summaries and things that you can find out.
There are tools out there I E gong
that have been doing this for a while.
They're very established.
But if you don't want to go spend $13,000 to $15,000
to try it and see if it works, it might be
a good option to build it yourself and see. Right.
So keep that in mind as you set up your AI committee.
You start talking about these things.
Those are some other things to consider as you
go looking for solutions or consider building your own.
So let's jump into headlines.
We've got a couple of really interesting ones here.
Per usual, Microsoft AI researchers accidentally
exposed terabytes of internal sensitive data.
All right, give me an idea what's a terabyte?
Terabyte is 1000GB okay, so multiple terabytes.
It says terabytes.
I'm inclined to be like, well, that's not a
terabyte is big, but it could be worse.
I mean, it could be worse.
It could be I don't know what kind of data this is.
There is some sensitive information like
passwords, I think two entire computers
worth of information it's exposed.
Essentially what it came down to was something in here
was misconfigured to allow full control rather than read only
permissions, which is kind of a big deal.
But then the article kind of wraps up
with like, this is just something that we're
all going to have to deal with. Deal with? Yeah.
I don't know what's your take on mean?
It says these are becoming
increasingly hard to monitor.
And mean, we have looked at Microsoft Azure
as a platform in our AI strategy with
I mean, does this give you pause? What do we do next?
How do we mitigate similar risks?
I think if you're going to be utilizing large
sets of data to train large language models, this
is probably a really big risk for you.
But as I read this, it talks
about the data also contained other sensitive
personal information like passwords and secret keys.
You're probably not going to train your
data model on passwords and secret keys.
So part of me really wonders, were these
people in dual roles, did they have access
to more information than they were supposed to?
Kind of goes back to the process
side of the security aspect here.
I'm also really curious what kind of
computer that they were running that they
had 38 terabytes on their personal backup.
Either that's all on their computer
or sounds like someone in marketing.
We got all these video files. Who knows?
We had to get our own server here at
Ben just to house all of our files. Well, totally.
I mean, it's not as big as it
used to sound like, but it's definitely like
a question of, well, what 38 terabytes?
That's 38,000gb.
This is just a ton of information.
And so one of the things that is interesting is to train
an AI, you have to have lots and lots of data to
be able to train it on all of these things.
So it really will continue to be a growing issue.
But it really goes back to the process
side of that people process and technology piece. Right.
They had misconfigured it.
Well, if they misconfigured it, then
they probably didn't follow the process.
I think in general, my question around this
is obviously what can we learn from it?
If Microsoft is having this issue and they're a big
tech company, is this just kind of an anomaly? Really?
It's probably just make sure that whatever
you're doing, especially if you're feeding it
data that you would consider sensitive, that
you have security controls in place.
Yeah, security controls and processes.
And then one of the things that I see over and
over again in these security checks is that there are audits.
When was the last audit that you ran?
What were the results?
And then if you're in big tech,
you probably are used to this, that
somebody is checking and performing audits.
And in a small business, you just don't
really have enough time to really do that.
So kind of instituting processes and policies and
then at the very least, having something that's
automated that does some kind of check would
have been really helpful here.
Like when people get busted for downloading a file
and they're trying to hack it to their competitor,
there's something that checked to saw that the file
was downloaded, like some kind of event popped up.
Something like that can help mitigate risk like this.
Yeah.
So put the tools aside and think to yourself, what
are those things that we wouldn't want someone else to
have access to and build process around that.
That's why our salesforce starts out with it depends
on how you think about it, the most security
or the least amount of roles and responsibilities.
And that's why Brent complains all the time, because
you only get what you need to know. Right?
Or you need to have access on a need to know basis.
He's definitely on a need to know basis.
Oh, poor Brent.
Shout out.
Okay, well, that was a good one.
Well, this next headline, we're taking a 180 here.
Austin church holds AI generated
service using chat GPT. Interesting. Yeah.
So this pastor out of Austin, the Violet
Crown City Church in North Austin, hosted a
Sunday service entirely created by AI.
So he had been hearing a lot about it.
He has some software developers that
are members of his congregation.
And he thought, well, why not?
Let's see what it can kick out.
And he said it produced about a 15 minutes service.
And not surprisingly, his observation was that
it was still missing the human element. Right.
And then it further caused him to contemplate.
I think this is a good question.
Kind of like what is sacred?
AI can't actually express, at least in its current
state, emotions of love and kindness and empathy commonplace
or what you would expect in a church.
So I think overall, I applaud him for trying the tool.
We keep saying, unless you play around with these tools and
just try to understand them, then how can you actually what
did you say when I sent this to you in Slack?
You said something.
Mark my words.
Mark my words.
There's going to be a religion.
It's going to be a large language model. Religion. Yeah.
Yahweh. AI. I don't know. I'm not the marketer.
I can't come up with the name.
It's like Christianity.
Yay.
I don't know.
You need to come up with the name
of the religion, because I would just fail.
But I could totally program it to take on some
core tenet principles and then it could write some gospels.
I do think this will be very interesting.
It sounds more like they had a fun idea and they
executed on it and it didn't turn out so well.
But I like the take.
I can guarantee you there's going to be some
kind of weird religious thing that pops up some
point and they're going to try to basically form
an organization and make money, but avoid tax.
You heard it here first, folks.
I do think that there's probably some people out
there that have used it in a way similar
to if you were to go to Google and
search on verses about grief or something like that.
And maybe it could be used to help build
thoughtful texts or emails that maybe you just can't
find the words for rooted in scripture.
So that's not really a revolutionary use case.
But I do applaud the pastor for trying.
I wonder if they streamed it.
We should go check it out.
Oh, that's a good idea.
We'll do a follow up segment. Yeah.
Chase and Mel go to church.
AI church. AI church. AI church.
Oh, there we go.
See, that was a good I don't know.
I don't know what you were going for earlier.
Well, I think we've covered a
lot of ground today, per usual.
Thanks for tuning in.
We want to hear from you.
Thanks, Jason, for writing in.
If you have a question or a comment, a
take on this or anything else that we've covered
up until now, there's something that we're not talking
about that you think we should be talking about,
please send us an email to thejunction@ventechnology.com.
Until then, keep it automated.
