#1: What CFOs Need to Know About AI & Automation

Welcome to the first episode of The Junction,

the podcast that explores the intersection of people,

process, and technology for SMBs and nonprofits.

Brought to you by Venn Technology.

I am your host, Mel, and I'm joined by my

co-host and enthusiast, among all things AI and automation, Chase.

In today's episode, we're exploring ways

AI and automation are transforming finance functions

and why CFOs should be paying attention.

I think this is something that

we should be pretty comfortable with.

Chase, we work with CFOs all

the time to automate their business.

I'm just kind of interested in your take

in general, in how AI is changing the

space and changing your conversations with CFOs today.

A lot of what we do is focused on automating

so we can get back to what we're good at.

We say that all the time.

And if you go dig into that statement of

getting back to what you're good at, it's really

taking out the monotony of the work so you

can focus on maybe a little bit, being a

little more creative in what you were doing right?

Having more focused insights, instead of the monotony of

typing into the keyboard over and over again.

The same thing.

If it is a robotic type process, that's

something that should be automated so you can

literally get back to what you're good at.

Let's think about the finance

professional today, the CFO.

We've seen their role evolving in

the last, even just year. Right.

Talked to a CFO last week, one of

our clients, that he is over procurement. It.

There's like a supply chain element to it as well.

Oh, yeah.

Their role has evolved so much, and they're making

decisions now, not just based off monthly reports and

the books, and they're trying to keep that going.

They're also getting brought into

these big digital transformations.

So, as you know, we held this webinar with

CFOs trying to understand how is this impacting you?

Are you using AI.

And we had some awesome speakers on to talk about it,

but we led the conversation with a poll and more than

half was like 60% said they're not using AI. Right.

Now I'm probably not surprised, but is that kind of

what you're seeing too, especially with this role type?

I think a lot of the business world is

still struggling to even get their processes written down.

We've had such a massive influx of data, like,

I don't know, go back 20 years, and we're

working in simple emails, simple spreadsheets, and now we've

got hundreds, if not thousands of database tables.

You just throw way more data at it,

and now it's a lot more challenging.

And so when you tack on the next

layer of that, right now we've got really

complex data models, really complex data architecture, infrastructure.

Then we throw something else on top of that.

People are still kind of

mastering the level below that.

And now we've thrown something else

on top of it, right?

These guys are really struggling to get that down

and that's why we've focused on that piece.

And we're now looking at how do we use

AI to enhance the automation that we're doing.

Yeah, no, you're right.

It's good to level set there because like I said,

even the same client I talked to last week, if

they didn't have an integration between systems, we're not talking

AI, just APIs one system to another.

They would be handkeying in information from one system

into another so that then their other functional team

could take that and go book the installation.

So you're right.

We're sitting here talking about AI.

We've got clients out there trying to grapple with

their process today and they're still handking information.

And let's be clear, we are not immune from our own.

We're a growing scaling company.

We're still figuring out and mapping our processes.

And as we add technology, it's so easy to go sign

up for another SaaS app like, hey, I'm going to go

add this thing to my marketing that just came out.

What, last night? Yes.

Let's just go sign up for it. Yes.

You tack in all of these

things that are constantly changing.

You adapt your process to that and now all the

things that you set up yesterday maybe aren't useless, but

they don't automatically adapt to what you were doing.

So now you have to go change those things. This process.

Everybody knows change management, right?

But doing that in your day to day work is way

more challenging than it is just to talk about it.

Like, sure, just change the way you do it.

Well, if you've got multiple departments moving data

by hand and you've got automations and integrations,

doing that automatically one simple change could mean

hours and hours of work.

So it's just a lot more challenging in today's

world to try to keep up with the pace

of technology, the ever changing process for your own

company and then keeping the whole team in alignment.

We really struggle with that internally.

Keeping everybody on the same page is a challenge.

You take all the automation and technology out

of it and you've got this core issue

of communication and that can be really challenging.

So let's take it back to our CFOs.

So we're grappling with how do we deal with this?

Let's say we have our processes in place or we don't.

Because you're saying right, as these tools come out,

we maybe adapt and continue to adapt our process.

Some of the threats that were identified

in the webinar, job displacement, staffing shortages,

I keep hearing this one, like it's

hard to find people for their teams.

But then I would think, well,

can't technology maybe fill that gap? I don't know.

Embedded bias, cybersecurity threats, data privacy.

Also, we're dealing with the books of

the financial performance of a business.

Like, there has to be a lot of gates around that too.

You don't just want to go upload

your income statement into Chat GPT, right?

Yeah, I mean, if you could do that, it'd be great.

Just tell me all the answers.

And if you did that, if you're someone

out there and you're listening, let us know.

No, don't tell anyone.

Well, don't tell anybody, but then email us

and let us know how you did it. Yeah.

I think as a CFO, one of the things that if I was

the CFO, I would be focused on, well, a human can inherently do

this better out of the gate than an AI model could.

So I would be focusing on nailing down the process

and ensuring that we are doing it maybe not like

100% efficiently, but as efficient as we could get it.

Because the moment that you add in automation and

then you tack in the AI layer, you're just

ingraining or maybe say I'll say dig a hole.

Like you're digging a hole into that process.

Not to say that it won't change, right?

It's a lot easier to tell, hey

Mel, can you change that real quick? Right?

It's a lot harder to tell an automation

or an AI, hey, I switched the fields.

Know, typically you have to bring other people in.

So I would focus on let's get that process nailed down.

And if I can't effectively tell somebody, an individual how

to do this well, and if they can't do it

well, the AI isn't going to be this magic bullet

that's going to step out and be like, boom, I

got it, and kick out this human person.

So when we're talking about start with your process,

what's an easy place is it time savings.

You kind of go do an audit of what

takes my team the most time, what's most painful

is it a blend of that and our business

objectives and we need to get there faster.

Pragmatically, where should they start?

By identifying those opportunities or

the processes that don't exist?

Just make a list, write it down.

I think it's probably the things that are most

time intensive that potentially offer the most value.

Now the value could be like, well,

we got back 10 hours a week.

Or one of the things that we see with some of

our clients is, well, we were charging for 10 hours.

Well now that happens in ten minutes, but I

still want to charge for 10 hours, right?

So just charge them ten times $100

an hour, whatever that is, right?

Just charge them $1,000.

So those are the things that

are the easiest to pick out.

And that's really focused on

the automation side of things.

The AI piece is more insightful.

Let me glean information from what is happening.

You take that 10 hours of automation, right?

And then you put layer AI on top of that.

And now I'm getting even more value.

So I would focus more like let's automate stuff

and then let's layer in the AI, some of

the opportunities that were identified in this same webinar

I keep going back to again, we had CFOs,

acting CFOs on the webinar with us, and then

we had some participation from the audience.

We asked, what opportunities are you seeing?

And there were kind of some use

cases in the risk management, risk reduction

kind of area, staying in compliance.

There was actually even customer retention, which I was

kind of surprised to hear from that group.

And then staff retention or augmentation.

So we had this interesting

dichotomy of staffing shortages.

And then also, well, no, we can retain them using AI.

And I thought that was really interesting because

there's a lot of people out there wondering,

is AI going to take my job?

Is AI going to replace me? Right.

What's really interesting about this, and one of

the ways that I've started to think about

it, is that these large language models, and

this might be oversimplifying it, but they effectively

amplify things that we are already doing. Right.

We generate content, we have these HR processes.

We do these things manually.

The AI can automate that and then glean the

same types of intelligence that we do as individuals.

So you take every facet of what somebody has done in

society, and now you can amplify that by 1020 X, right?

Like, we've got cybersecurity issues.

Well, people are manually pinging your

computer, trying to hack in.

Well, now I can do that 20 times the rate,

and I can do it 24 hours a day.

They were already kind of doing that, of course. Sure.

But now thinking about generating content, right.

Quality, mostly quality content is written by hand.

Well, I can train an AI model to write like you,

and now I can do it 24 hours a day.

Well, naturally not. Right.

But you have this idea of amplification

of the things that we're already doing.

Some of those are great things.

Some of those are like really ethical questions. Right?

Like, well, are we going to replace Mel with

this Melbot and start writing a bunch of content?

You better give her a cooler name if

you're going to do Bell MelBell Autobot.

It's like Transformers. I'm with you.

There's some really we've been getting

our hands on it here, right?

I've definitely been seeing the light.

I'm the first to say that

I was very apprehensive about it.

And it came from the fear of the unknown. Right.

And then once it was like, you

actually kind of gave me that nudge.

Hey, go play with it.

What if you type this in?

You even sat side by side with me and

showed me how to build a website using GPT.

It's possible, right?

So I definitely see it.

It's not going away.

We have to address how we can use you know, I'm thinking

again about the CFO and kind of so one of the things

that Kenny Mullikin, he's actually a CIO, who was on our panel

said what he was excited about was the ability to put tools

in place to help people do more mean.

It's exactly what we're talking about when we talk about

integrations and automations, you're talking about it more as layering

the AI on top of that for that additional insight,

he also pointed to that as well.

But anything that we can do in our own

orgs or for our clients to provide teams with

the information that they need faster, or again, the

insights, because there's things as I've been playing with

it, I'm starting to find myself asking questions that

maybe I wouldn't have asked before, right?

It's almost like a little personal assistant.

You're kind of like throwing stuff out there to see

what sticks and then the answer causes you to go,

okay, well, let's dig deeper on that, right?

Well, what's really interesting, especially if you've

been on Chat GPT, there's some other

models that you can play with.

You start to ask it more questions, or

you start to ask your questions differently.

And what I've found is that these large

language models feel a lot like, I don't

know, a young teenager, like an intern, right?

Well, you're not going to ask your intern like these

really complicated questions because they don't know or they're going

to give you a really dumb answer, right.

But these large language models have somewhat

of a really good insight, right?

Because they've read the entire Internet

and aside from hallucinations, they can

respond in an intelligent way.

And if you think about Chat GBT just like

literally being an intern, well, you're just naturally not

going to ask that person certain questions, right?

Because they just don't know enough or

they haven't been there long enough, or

they're in a different department.

You can ask these things a wide range of

questions and they start returning, I'll say for the

most part, like genuine, authentic, real answers.

I don't want to dismiss the hallucinations because they are

there, but ultimately it amplifies and gives you a lot

more access to knowledge that you would otherwise neither not

know or have to go look and find on Google

and spend an know, digesting all of that content right?

Before we get too ahead of ourselves, I

actually was talking about this over the weekend

with some friends, talking about Chach Abt hallucinations.

I said, what's that?

What's a hallucination?

So I think it's important we probably define that.

So for anyone out there listening

that doesn't know what is it?

Some people maybe just naturally do this, maybe it's

I won't necessarily call it unethical, but when somebody

asks you a question and you're like, I'm like

80% confident, but you go in 100%, right?

And then you're wrong, you were wrong, but nobody calls

you out because it's a social setting, we're worried about

it later, or they have no idea, right.

And they just take you at 100% right

like when I was explaining to my friends

what Hallucinations was, that nobody knew.

So they were like, oh, that makes sense.

Yeah, I'm confident I delivered the answer to

the best of my ability understanding, because I've

learned from you and others in this space.

But yeah, so it's basically we need to be aware

as we continue to use these tools that it's trying

very hard to give us an answer with great confidence.

Yes, well, you run into other things.

The easiest thing is you can tell these large

language models that one plus one equals three and

then it will say no, it equals two and

you can say no, it equals three.

And then you can ask it again what is one plus one?

And it'll say three.

Like it has this mentality or this want and desire

to effectively kind of please you and it will try

to do anything it can really, to achieve that goal.

You don't really see that when you're

typing it out in a chat format.

But as you train these models on people and content that

are trying to answer your question as the best way they

can, these models will naturally do the same thing.

So they naturally overconfidently answer your

question, even if they're wrong.

Sure, yeah, it's good to be aware of.

So some other use cases, kind of taking it back to

what we were talking about with our CFO audience, something as

simple as having it write your notes for a board meeting

or putting together kind of a deck or an outline.

I've heard so many, I mean, I'm surprised by the

number of just C suite executives saying, yeah, I'm actually

using it to help me just kind of like craft

an email, saving me so much time.

And then I can say, hey, make it more

formal, or make it more one of your favorite

words, authentic, less formal, more conversational, witty, insightful.

So I think don't sleep on the email. Right?

So if you haven't started playing with that,

maybe start there and see how it can

help expedite the time that you spend sitting

there trying to craft an email or again

summarizing the financial performance for the quarter.

For the board, you can give it some parameters around

how many words and kind of give it some structure.

That's one interesting use case, but easy.

I think anyone can try that.

There's also, I've seen typing in you

mentioned using it as a Google search.

So what are the gap rules for this?

Or what's the code for?

I from what I've seen, we had another client of ours

that was using it for that and he said, man, I'd

have to go out there and search Google and try to

find the right exact code or compliance to site and chat.

GPT delivered it to me in 2 seconds.

Yeah, so even things like that, you go

back to Hallucinations and the only thing you

have to really be careful about.

And you have to do this with people too.

You have to trust but verify.

Like, nine times out of ten, Mel, you're right nine

times, maybe ten times out of ten I'm right.

But that one time I'm wrong.

And if I rely on that information to be

100% right all the time, well, I'm in trouble.

These large language models are the same way.

You have to trust but verify.

And if you're relying on the content that it delivers

to then go and effectively risk your job on the

content that it's generating and you're not verifying it, then

maybe you're going to get what's coming to you.

But if you go in and verify what it's saying and you

copy and paste some of that into Google search and you can

validate that, then great, you saved a bunch of time.

Yeah.

All right, let's move on to

our last segment of the podcast.

We're going to talk about headlines.

AI in the news, so pulled up a couple here.

56% of business leaders are

incorporating AI into cybersecurity.

If you remember earlier in the episode, we talked about

how cybersecurity was one of those kind of opportunities or

threats, depending on which way you look at it.

And I think, again, CFOs, this remains

an increasingly important topic as businesses continue

to embrace new technology or platforms.

So I'm just kind of interested in your thoughts

on that, especially just given everything that you're seeing

and there's a lot of excitement around how we

can go use it to do more.

But I do feel like I have not

heard and I've not personally went and sought

out, what are the security threats?

What should I be concerned about

outside of what you've shared? Sure.

I'm going to oversimplify again for the laymans out there

that the data that they're looking at are the logs.

Like when a hacker is trying to hack

into your stuff, logs are getting generated.

It's a database.

It has tables, it has columns or fields.

And the AI is looking at those data

points to determine, is there an anomaly here? Right.

Like, every time you hit our network, we

can see that Mel's IP hit this.

Well, if we get something from out

of the country, well, that's an anomaly.

Those are things that we should start looking into.

And that's what cybersecurity experts when cybersecurity first

came around, that's what they were doing.

They were looking at the logs.

Wow, this log is from China.

We don't have anybody in China.

And so effectively, AI is kind of doing the same thing.

It's analyzing the data to determine where those anomalies

are so it can call those out and say,

hey, we've had a huge know those Russian crazy

people are hitting our network really hard.

Call that out to those experts.

So they can then go either block the IPS or do what

they need to do to ensure that the threat is minimized.

So we're going to see a ton of that.

Whether it's like the software that's being

generated includes these natural AI elements.

I mean, you can really just attach AI to the

databases where these logs are being generated, and there's certainly

way more complexity that goes into all of this.

But at the end of the day, these AI models are

looking at the data to determine where that risk is and

helping you mitigate it by bringing it to your attention.

Right.

So instead of kind of, again, looking at it,

we're more glass half full folks over here. Right.

So certainly it could be a threat, but

it can also be used as an opportunity

to help further quickly identify those threats.

Oh, yeah.

I mean, from a cybersecurity standpoint,

you're always trying to identify anomalies.

And so for business leaders, we want to minimize any

kind of risk that we have from data leakage.

Everything that we do on our network generates some

sort of log or some sort of data point.

And so if we can use AI to speed up that

identification of that problem, we can knock that out, we can

close that leak, we can derisk or minimize the risk.

A lot of these things that we see in

the news are from people just not knowing, right?

They have no idea.

And then it goes on for months.

Six months later, they finally figure out that, well,

the Russians have been in here for years.

They have no idea.

And so with AI, now you have this

ability to see what's going on on your

network faster because the AI doesn't sleep.

It's working 24 hours a day.

It already is impacting cybersecurity because

cybersecurity is so focused on data. Right.

They're looking at the data all

right, onto the next headline.

With AI, top CFOs are more likely

to see it as a job destroyer.

This is one of those things where you can look

at either side of the apple, the pie, the orange.

Yes, it will destroy some jobs, but it's going

to do so much more than destroy jobs.

It's going to create jobs, it's going to make people faster,

better at what they're doing, and it's going to eliminate areas

where we don't need people to do the types of jobs

that AI is going to be taking over.

Think about data entry.

Nobody is waking up in the morning be like,

I can't wait to get my Excel sheet and

start typing in stuff like, let's do it.

Some people are really attached to their Excel.

If you want to do that manual entry, you let me know.

That's not me.

I'm just saying, okay, but you have these jobs

that really, I'll say, like push paper, right?

People are pushing paper.

Those jobs are going to go away because

AI is going to take that over. Well.

Now those people can do something that

is more important than pushing paper.

So yes, it's going to destroy jobs.

But I think what's more important here is it's going

to, like I said, amplify what we're already doing.

And if you can take that in and digest it,

then you're going to understand the power of this.

It's going to help you code better, it's

going to help you write content better.

All of these things, if they're used in an

ethical way, are going to enhance our productivity levels

to a place where we've never seen them before.

All right, so let's move into a hot take.

I know that was kind of already your hot take.

So that's hot take one.

I have a follow up to that.

So do you think that businesses are

responsible for providing alternative employment opportunities or

retraining programs for employees where they are

replacing them with automation?

What is the responsibility?

We see some of this already and even without AI

in the automation that we do and the intent always

goes back to getting people to what they're good at.

So do they have a responsibility to effectively

retrain or reprioritize what these people are doing?

I would say in a sense.

And not that they're morally obligated, but that

they're naturally going to find things for these

people to do because they already have the

institutional knowledge to take on the next thing.

Right.

If they're going to let that person go and

then find somebody else to do another gig, well,

they just wasted a whole bunch of time.

So I'm not too concerned about this.

Other than the places that have hundreds

upon hundreds of paper pushing people. Sure. Yeah.

Most of the folks that we're working with,

they didn't come in especially into more like

executive positions to do more manual tasks. Right.

They were brought in to focus

on strategy and leading the business. Right.

So they're doing things that they

weren't actually hired to do.

That's what ends up happening. Right.

Well, and above and beyond that, if you think of

an AI as that intern or this new level of

workforce, you're now kind of responsible for those people.

People. Right.

They're generating and when I say they, I mean the AI.

Right, whatever you're using AI for, you're

now responsible for what it's doing.

You don't just get to check out and call it a

day because you have to see and oversee what's going on.

Because if you don't and you're risking your job on that,

well, maybe you're going to be out of a job.

So there's going to be other things that come

out of this that are going to be responsibilities

that these folks will have to take on. Awesome.

All right, well that's a wrap for today's episode.

Really hope you guys were able to

gain some valuable insights on the ways

AI and automation are transforming finance functions.

Obviously we kind of bobbed and weave

into some different parts of the organization.

But again, as technology keeps evolving, SMBs nonprofits

really, all of us need to stay on

top of the game and embrace these new

advancements if we want to remain competitive, right?

We want to know what questions or thoughts you have.

Email us your take at thejunction@ventechnology.com.

Thanks for tuning in.

And until next time, keep exploring, stay

curious, and embrace the power of AI.

Because it is here to stay. Keep it automated.