The Accounting Technology Lab

Brian F. Tankersley, CPA.CITP, CGMA: Welcome
to the Accounting Technology

Lab, brought to you by CPA
Practice Advisor, with your host

Randy Johnston and Brian
Tankersley.

Randy Johnston: Welcome, the
Accounting Technology Lab. We're

coming to you live from Orlando
in the Marriott World Center,

where we're attending Scaling
New Heights 2026 Now it turns

out Brian and I are fortunate
enough that we've produced

sessions here, but just wanted
to give you a sense of what's

going on here, and as you try to
evaluate which show or shows

might be appropriate for you,
you know, this year's show

actually had quite a little bit
of attraction, but it was called

Strange New world, and you know
there's just a lot of things

that are strange with AI, but
what wasn't strange is

attendance was solid, about 2000
attendees here, and by my count,

133 exhibitors. I had the good
fortune to actually do a walking

tour of all of the new
AI-powered general ledgers,

which we've covered in other
accounting technology labs. I'm

sure there's going to be more
that we'll talk about in the

future, but just to set this up,
the Sunday sessions, the

orientations for first-time
attendees, I actually decided to

attend for a change, and it was
healthy because I got to hear

what newbies were being told,
and I don't have a good count on

the first time attendees, but I
would say it was probably 100 to

150 and it might have been more
than that, and then the practice

advancement session, which Joe
has taught frequently on Sunday

mornings I had not sat in one on
a while, and he did a nice job

talking about AI, and Joe
basically started the 40,000

foot overview of AI, and then
drilled down to a 30,000 foot

overview, and then down to
20,000 and you've heard me do

that. In fact, we've both done
that, talking about workflow

through the years, but Joe's
points were that, you know,

here's the big picture on AI
with all of the Chat GPTs and so

forth, but the landscape has
really combined has evolved into

a combination of the large
language models, the machine

learning, and so forth, and that
the 20,000 foot view says that

you can actually use AI to
provide more advisory services,

I think that's probably true, by
the way, and then at the 10,000

foot view, actually having AI as
a growth engine, and then what

he would call on the ground,
where there were actual live

steps, overall a nice setup for
the people that were there, and

that is a lot of the members,
and a lot of the newbies got to

hear all of that.

Brian F. Tankersley, CPA.CITP, CGMA: Unfortunately,
instead of being in fabulous

Orlando, I had the privilege of
spending the day in the Miami

airport, courtesy my friends in
American Airlines, but the good

news is they had an Amex lounge,
so you know, I noshed and

watched World Cup, and just had
a good time, but unfortunately

he only had the main man here.

Randy Johnston: Yeah, so on that
point, then I attended a couple

of sessions. I wanted to see
what Mara Negarovsky said about

bigger data, and that was a nice
session. So again, we know that

to make AI work properly, you
have to get your data clean, and

we've talked about that, and
we've seen that in other events.

Brian F. Tankersley, CPA.CITP, CGMA: I
like his operation, he just Nuba

Networks and Negroski
Consulting, I mean, it's he, for

so many years, has solved so
many pain points in CAS that

nobody else did.

Randy Johnston: Yeah, and when
you look at that whole

Mendelssohn operation, you know
he's just trying to do a lot of

the right things, but you know,
like we learned when we did the

black ore release work in New
York, everybody to a person was

talking about cleaning up data
consistently here so far, you

know, cleaning up data is still
a big deal, and it was clear

Mario was on that same path.

Brian F. Tankersley, CPA.CITP, CGMA: Well,
you can't make filet mignon at a

hamburger. I mean, it's, you
know, and if you're gonna, if

you're gonna make filets, you
gotta have good cuts.

Randy Johnston: Yeah, that is
absolutely true. So then I was

fortunate enough to see Christy
Monahan do a standard operating

procedure written with AI
sessions. She again did a nice

job, and it kind of set the tone
for, you know, the way I was

thinking about a lot of the
sessions. Usually at Scaling New

Heights, the sessions tend to be
rather short and practical, and

all of them are recorded so
members can see all of the

sessions, so unlike us, where we
maybe just get to see a few and

present a few, that's a little
bit different, but it was such a

pleasure then to do Tech Update
here, you know, the crowd here,

we've got a long-term
relationship with, and we just

had fun doing Tech Update, and
you've seen me do Tech Eight

update for fun, you know, a few
times. Times this year, but

there's a lot of content that is
the hardware needs and the AI,

and again, just a good overview,
but it really, I think, just

resonated well with the crowd
this time. So that got us over

then to the main stage opening
events, so that usually happens

for Scaling New Heights on
Sunday night, and the way Joe

tries to manage the show,
there's usually main stage

events to kind of end the day,
start the day, so book in the

day, if you will, and he had
brought in a well known as the

world leading tech magician,
Akilan Leiser, out of the UK,

and his wife, I think her name
was Elizabeth, but you know,

bottom line was he did some
wonderful tricks for using iPads

and robots, and so forth. And
what I thought was fascinating

about that, and you didn't get
this experience, they said they

would not open the doors once
they started the main stage for

at least 10 minutes until the
first sequence was done.

Brian F. Tankersley, CPA.CITP, CGMA: That
sounds like something you do on

Broadway when you have, you
know, like some monolog or

something,

Randy Johnston: and it was
absolutely didn't need the

interruption, so and he did a
really nice job, but you know,

being able to have an iPad where
you know it has a picture of a

tennis ball and you pull the
tennis ball out, as an example,

or a spinning wheel, and then
all of a sudden he's got a

medallion in his hand, or one of
my favorites was the

butterflies, where the
butterflies started coming out,

and all of a sudden there's
butterflies flying everywhere,

or the reverse, it had a red
scarf, and it got sucked into

one of the iPads, that type of
thing, and they used a robot

dog. Well,

Brian F. Tankersley, CPA.CITP, CGMA: some
iPads do suck.

Randy Johnston: Yeah, well, that
does happen, but this one was

hard to picture where it went,
but the robot dog, where they

actually did card tricks, was
also fascinating, because the

robot dog picked the card that
was identified by an audience

member, written on by an
audience member. You know how

magicians work, it's kind of
deal, but that was kind of the

idea, and it turns out that Joe
opened up with just a few

general comments for the
conference, and then did an

interview with Sasha Orloff of
Puzzle, and Sasha's company,

Puzzle, now has about 10,000
users, which is good, and they

just chatted about how AI can
serve the profession, and it

should not be the other way
around, was part of their

conclusions on that, and also an
interesting observation. The

question about AI replacing
accountants, no, but 50% of all

token spend right now, they
said, was for software engineers

and the salaries were the
highest they'd ever been, so

not.

Brian F. Tankersley, CPA.CITP, CGMA: And
I have to say this about the

token spend, I think that's an
important thing when you see the

stats, because when you look at
the stats on subscriptions,

people that are buying this
through subscriptions, Claude is

only like five or 6% but
everybody that's developing, or

most of the people developing
software, either have a $200 a

month cloud pro plus titanium, I
don't know what it's called, but

some superlative plan that has a
bunch of, as a bunch of tokens

available, but most many people
are just buying tokens, kind of

like you buy gas for your car
and then use them, so they,

they're not. It's harder to
measure who's using how many

tokens and what the real user
count is.

Randy Johnston: Yeah, they're
very difficult to figure out.

So, you know, then at that
point, you know, a little bit of

a history walk with Joe, but
this is the 18th scaling new

heights, and he started the
first one at the Crowne Plaza in

Atlanta with 250 people, for
this to be 2000 people this time

around, you know that's quite an
accomplishment. But next up was

David Barrett, you know, David
founded Expensify, and actually

something I didn't know, and I
can't verify, in fact, but I am

going to repeat it here, David
Barrett had coded Expensify with

blockchain the year before
Bitcoin's blockchain came out,

Brian F. Tankersley, CPA.CITP, CGMA: really,

Randy Johnston: and I, that's
pretty verifiable, and it's like

that's pretty interesting, you
know. I knew he was, you know,

hardcore coder, but that was
also interesting, and then he

went on to explain how AI should
work and what the experience

should be like, and I hadn't
really thought about the AI

enablement in the expense
tracking products that he's

built, but he was very
thoughtful about how he deployed

the AI, and so that was worth
seeing, and then really the star

of the show was, you know, Dr.
Daniel Susskind. We've

recommended the book before. His
book originally was out about

2015 and I reread it
anticipating his presentation.

Brian F. Tankersley, CPA.CITP, CGMA: What
the professions.. what's it

called.. the, you know, the next
generation of professions, or

something like that,

Randy Johnston: and that. Book,
as it turns out, he's actually

got a new one coming in the
fall, which is about, you know,

what the, what you should do
with children, so forth, but the

key here is that he quoted a
fair bit of statistics and tried

to show the arc of what had
happened with agricultural and

manufacturing, and so forth,
and, you know, also the impact

on several different areas, so
he basically said, "Look, I'm

going to talk about six areas:
I'm going to talk about blue

collar, I'm going to talk about
white collar, then I'm going to

talk about AI, and then I'm
going to return to white collar,

and then I'm going to talk a bit
about education, and then

artificial general intelligence,
and it was a nice arc. He moved

fairly quickly. There were a lot
of facts that he covered, things

that you and I have talked about
in the past, and you know,

including things like Deep Blue,
the chess piece, and you know,

the Garry Kasparov defeat. And
then

Brian F. Tankersley, CPA.CITP, CGMA: 30
years ago, isn't it? Now, I

mean, it's

Randy Johnston: 1997 which is
hard to believe, and then also

the Jeopardy situation with
Watson, which he cited, and you

know, his father was actually
early on in AI and created the

first AI-based law system, so
father-son, or you know, both

heavy into this stuff, but the
disruption,

Brian F. Tankersley, CPA.CITP, CGMA: but
that's a fun Thanksgiving, isn't

Randy Johnston: it? Yeah, you
know, sometimes you can

Brian F. Tankersley, CPA.CITP, CGMA: get,

Randy Johnston: how do you say,
blah, yeah, kind of like

Brian F. Tankersley, CPA.CITP, CGMA: halbla
hubbla, hey, huh, blah blah

blah, you know,

Randy Johnston: kind of like
Charlie Brown and Lizzie,

exactly. but it was fascinating
because he cited an Economist

article from October 17 of 2015
which I had put in Tech Update

that year, and so it was a
little bit like Old Home Week,

but many of the statistics I
knew from his book, but also had

incorporated in our
presentations through the years,

so

Brian F. Tankersley, CPA.CITP, CGMA: future
of the professions, that's the

title,

Randy Johnston: future the
profession,

Brian F. Tankersley, CPA.CITP, CGMA: the
title, it's great book,

Randy Johnston: it is, so as it
turns out, in 2017 McKinsey

published a piece saying that 5%
of jobs can be automated, and he

breaks down tasks, and that's
really the key learning here.

His theory is that even if
you've got a profession, many of

the jobs are tasks that can be
automated, and so if you can

automate the tasks, that gets
you some leverage. But he went

on to say it's the human
interaction and the creativity

that the machines have not been
able to duplicate at this point,

and then he went on to talk
about the Khan Academy and the

Mika, a piece that had been
built, and honestly, for this on

this recording, the book that
I'm in the middle of is AI for

Good Josh Tarantiel book, and
it's a great book, but I don't

recommend most books until I've
finished them. There's three

sections, and the Khan Academy
was the first, and it's like,

oh, that's very good, and right
now we're in Cleveland Clinic

territory, but in the big
picture of things, he goes on to

drive that education is the
response to AI, if you educate

and do the change management,
educate, but several attendees

that I respect here, and several
innovators are saying that

they're using as much as 1/3 of
their time on education right

now. So, if you think about it,
and we'll just talk 40 hour

weeks, you know, if you're
spending 1012, hours on AI every

week, most of us don't have that
type of,

Brian F. Tankersley, CPA.CITP, CGMA: I'm
spending about 30, yes, because

it's it's one of those things
that is, you know, for what we

do, which is training and
development and consulting and

workflow tracking and all that,
it's just, it's you can either

run as fast as you can now or
you can take early retirement,

you know, and so we're, you
know, tying those shoes,

Randy Johnston: yeah. So it
turns out, then closing up with

artificial general intelligence,
and you know, just end to end a

well thought out presentation,
but we, he suggested that you

take AGI a lot more seriously,
but it's going to take a

combination of business leaders
and politicians and parents, and

it's a profound change, but that
I'll say caution about taking

that seriously is a big deal.
So, in any case, then from

there, you know, the conference
started unfolding, and you know,

again, every time there was a
main stage presentation, usually

the magician would come back out
and do some sort of routine at

the beginning of their end, so
Keelan and his wife, as the

assistant, did a number of
those, and that really was true

on. Each one now, because of
time constraints, we probably

will talk a bit more about
vendor sessions and other

observations in another
accounting technology lab, but

it is clear that there's much
going on, and the level of

scaling new heights is quite
different this year than it has

been in prior years,

Brian F. Tankersley, CPA.CITP, CGMA: but
we still got, I still got to

have my audience with the Pope
of pricing, Ron Baker, you know,

and it's, I got to see, I got to
see, you know, Joe and Caleb,

and got to meet a lot of new
people here. It's always fun to

go here, because they're. it's.
it's. I don't know, it's some

very nice crowd,

Randy Johnston: yeah, that good
people doing good things for a

lot of businesses, and let's
face it, Microsoft was a anchor

sponsor, and realistically, your
ability, my ability to deliver

AI sessions here was probably
enhanced because of that, I

think between us we've got five
AI sessions in addition to tech

updates, and I will

Brian F. Tankersley, CPA.CITP, CGMA: say
this: we're not the only AI

sessions here, and you know,
Microsoft, though I will say

this, I think people don't
under, you know, people getting

all excited about Claude, and
they got really excited about

ChatGPT, but what I think they
don't get is that Microsoft is

taking the compliance very
seriously, and I think it's, I

think they're the only ones that
I'm seeing, especially in light

of the logging that they started
to do with the new Claude,

that's that even enterprises
could opt out of. I think that,

you know, the Claude five that
they pulled Fable, I guess it

is. I'm, I think Microsoft has
got the right approach to use of

AI inside of companies, because
they, you know, they're just,

they're dotting the i's and
crossing the t's, and as a

profession, where when you don't
dot the i's and cross the t's,

bad things happen. I, you know,
it's a very, I think you need to

pay attention to that,

Randy Johnston: and you know,
along that line, this is a big

year of AI governance. Last year
we said you always start with a

policy, but now you've got only
got to try to restrict things

with governance, and I brand
reviewing a new AI policy that

we've created, but it's got 26
different governance-related

issues on AI platforms,
including all of the bigs,

Walters, Kluger, and Thomson
Reuters, and so forth, but even

smalls like Kanopy interfacing
with files, and we talk about

the reasons that happens, but as
I started digging into what you

have to think about as you use
different platforms, there's way

more there than I thought, and
so the simplistic AI policy that

I'd recommended in the past, it
just didn't stack up to today's

needs, and you're going to hear
us talking more and more about

the governance and coding and
all these things that you can do

with AI,

Brian F. Tankersley, CPA.CITP, CGMA: and
I've got a session that I'm

doing, actually tomorrow, as
we're recording this. It's

Monday night, I'm going to
record the.. I'm going to do the

session tomorrow afternoon, but
we're going to talk about the

governance issues in pretty good
detail, and one of the things

that we're going to talk about
are privacy policies in EULAs,

because EULA, of course, is an
end user license agreement. One

of the things that I think that
people don't realize is that in

many of these EULAs they
incorporate by reference the sub

processors' user license
agreements, and sometimes they

don't even tell you who that is,
they just say we have sub

processors, and by the way, they
have their own user license

agreements, and you're stuck
with theirs, and it's, it, you

know, this figuring out who's
really who really has access to

your data and who's using the
data for training, you know, I,

I've done a bunch of reading
lately about the impact on HIPAA

and FERP, and a lot, and you
know, all the PHI and PII rules,

and it's very unclear what's
permissible, but it's also clear

that there there's a lot of
stuff going on that, at a

minimum, is highly frowned upon.
Yes, of using data for things

that are not good,

Randy Johnston: and vendors at
this conference, Scaling New

Heights, and the prior
conference, AICPA engage, who

wanted us to look at their
products, you know. Normally,

the first thing I do is go look
at the privacy policy, and the

last one that I looked at today
said we used sub processors,

didn't name them, didn't say
anything, didn't know reference,

and it was like, oh, good, but
you know, let's take a couple of

other names, I am

Brian F. Tankersley, CPA.CITP, CGMA: Oz,
that's who the sub processor is.

Yeah, the guy behind the
curtain,

Randy Johnston: behind the
curtain, for sure. But you know,

I want to call it two other
things, which we're not really

trying to talk about in this
podcast, but just of note, the

Meta Ray band glasses. There was
a winner from Main Stage today,

they got a pair of those, and
you know, if you look at the

price. Privacy on meta ray
bands, it's not pretty, and

Brian F. Tankersley, CPA.CITP, CGMA: you
met meta and privacy are two

words that should not be in the
same sentence

Randy Johnston: most of the
time, not. And then the other

big announcement with the
worldwide developer conference

from Apple was the alliance
between Apple and Google, which

will unfold as the year goes on,
so when we look at the new

generation AI computers,
Google's the Google book, which

is not a Chromebook, it's an AI
machine, and Nvidia's machine.

I'm watching this processing
power at the edge, and you know

where the big players are
playing, and we'll try to get

some clarity for you, that for
you on those topics in future

accounting technology labs, but
I gotta tell you, it's not clear

to me right now, so I'm not sure
I could explain it to you.

Brian F. Tankersley, CPA.CITP, CGMA: My
gut is that accounting is, we

are so reliant on having stable
systems that this new crazy

stuff at the edge that's
experimental and everything, we

have two or three years before
it really starts affecting us,

and the other thing is that,
with from a regulatory

perspective, because we have all
this super private data, we have

to be grown-ups all the time,
and that's the hard thing, you

know, because the sales people,
a lot of them can do all kinds

of crazy things with data, and
they can put leads in there, and

everything else, because, you
know, when somebody puts their

data in a form and they hit
send, you can do just about

anything you want with it, but
it's a, you know, the thing that

you've got to understand that
sets the profession apart from,

you know, from the sales people
and from the marketers and all

the folks like that. The
marketers and sales people

aren't doing anything unethical,
but they just don't have the

legal requirements that you and
I do with PII, and that's why I

think Microsoft is has, I think
that's why I think Copilot

really has some legs, even
though there have been some

criticisms of the platform. I am
quite excited that it's starting

to include some of the cloud
models, so you can pick your

model and yet still be
indemnified, and all that, which

I think is glorious, and one
other piece that Sasha, CEO

Puzzle, said, which is also
landing pretty well with me,

probabilistic versus
deterministic, and if you think

about our legacy accounting
software products, you put in a

number, you got the same answer
all the time,

Randy Johnston: and with the
statistical models on AI,

they're not quite as predictable
on the outcome, and the

analogies that he used in
several different situations did

make sense.

Brian F. Tankersley, CPA.CITP, CGMA: Yeah,
one of them was one of them was,

do you want vending machine, you
want the accuracy of vending

machines or the accuracy of slot
machines? You know, do you want

something where every time you
put a quarter in you get a, you

get a bag of chips, or do you
want something where you put a

quarter in, and maybe you do,
and maybe you don't. It's a..

it's.. and that's.. I think
that's one of the hardest things

for a lot of accountants to
understand about AI, is that it

is very ish, you know. It's not
some.. you know, it's kind of

like when you know you could
have a huge.. you could have a

huge English class that had 800
people in a massive chemistry

lecture hall, and you could give
them all the exact same

education, the exact same text,
and the exact same assignments,

and just like with AI, you would
get 800 different answers, and

some of them would be out here
on the bell curve as A's, and

some of them will be up here as
C's, and some will be down here

as F's.

Randy Johnston: Yeah, it's we,
as you said multiple times, we

live in interesting times.

Brian F. Tankersley, CPA.CITP, CGMA: Yes,

Randy Johnston: so we'll be back
with some more interesting times

for you in the Accounting
Technology Lab. We appreciate

you listening in today, and
we'll talk to you again very

soon today.

Brian F. Tankersley, CPA.CITP, CGMA: Thank
you for sharing your time with

us. We'll be back next Saturday
with a new episode of the

Technology Lab from CPA Practice
Advisor. Have a great week.