Leading with AI: Fractional Leadership and Personalized Innovation with Adi Vaxman and Mark Latimer

0:00

welcome ladies and gentlemen to the AI training podcast I am very excited about the show today I have an incredible

0:06

guest Addie vaxman head of Sheba Consulting a fractional leadership

0:12

Services practice out of New Jersey Sheba has been doing this for over 30

0:17

years including uh roles in the SE Suite helping startups and established

0:22

companies she's just incredible I'm so excited to have her on the show I could go on and on about her background and

0:29

history but I am going to let the interview do the heavy lifting here so thank you and welcome thank you so much

0:36

for having me Mark appreciate it this is going to be a lot of fun I know that we

Understanding Fractional Leadership

0:42

connected recently through a mutual friend and your experience in fractional

0:47

leadership and AI I've got so many questions but maybe for to start for

0:52

those who don't know what is fractional leadership fractional leadership is really the practice of bringing in an

0:59

experien se- level executive on a part-time basis basically to Outsource

1:05

that rule but it's not traditional Outsourcing because that person is usually 100% involved in the business

1:12

they are just like any other SE Suite executive except they're doing this part time and they're usually holding several

1:18

clients so our fractual leaders usually have a load between three and five clients depending on the level of

1:24

Engagement it's a very costeffective model um we do Co chro and CTO

1:30

and it's a very cost- effective model because you get a lot of experience especially that comes with the fractional model because fractionals see

1:37

a lot of businesses rather than one or two or three that normal people in their careers see and you get someone witho

1:43

many years of experience you don't have to pay what you would otherwise otherwise an executive at the same rate

1:49

would cost you I don't know anywhere from 350 to 600,000 a year or something ridic like that amazing that's awesome

Industries and Commonalities in Fractional Leadership

1:58

so I know you have a ton of EXP experience and I'm curious what industries do you tend to gravitate

2:05

towards we started out with the tech industry because both my partner and I come from the tech industry before we

2:11

started the company we were in the tech industry for several couple decades I'm

2:16

aging myself here and we expanded to several other Industries we currently

2:22

work obviously on Tech and anything tech software I think technology information and security we also work with a lot of

2:29

business services companies agencies marketing agencies market research agencies software development agencies

2:35

anything that's like a business service we work with Medical Practice groups that's been a niche that we've been in

2:42

for the past since 2019 so it's going on five years that have their unique

2:47

challenges and we also work with media publishing and some different types of

2:53

nonprofit just because of my kind of personal abilities to to take up causes

2:59

and run with them very cool and through this kind of

3:04

it sounds like quite a large mix of industries that you work with what tend to be the the commonalities between all

Challenges and Solutions in Business Operations

3:10

these companies most of them are in some state of transition so they are either

3:16

growing or they have grown significantly they have some kind of a challenge they need help with something some of them

3:22

are coming to us because they've grown too fast and they really can't handle it and everything's falling apart so

3:28

unfortunately that's like the majority of them that are coming to us when things are falling apart and they're

3:34

drowning and a small minority the smart minority is coming to us before that

3:39

happens when they know that there's a change that needs to happen when they're looking to expand when they're looking to grow when they feel that some area of

3:46

the business isn't functioning correctly they feel that things are not being very efficient or that they can't have you

3:53

know they they don't have enough control over what's happening in the business and many business owners especially in a

3:59

lot of the industry that I mentioned are their talents are in their industry right they know how to be a doctor they

4:05

know how to be a lawyer they know how to be a marketing person they don't necessarily know or want to know or care

4:11

to deal with all the operations and the behind the scenes of running business and they're doing it because they don't

4:17

have a choice and it's taking up too much of their time and they're looking for someone they can really trust to do

4:24

that and be able to give that away and trust someone else and that's us got it

AI Implementation in Business Operations

4:30

and as it relates to operations obviously over the last year especially AI has really taken hold of operations

4:38

what are some of the challenges that have come up with your clients in leveraging this technology since

4:44

November 2022 we've been we are all we're a very technological team in general we're all geek to put it plainly

4:51

so we love new technology whenever something new comes up we jump in and try to figure it out and that's what

4:56

happened in November 2022 when Chad GPT came out like there were earlier versions but they were not great like

5:04

we've been experimenting with AI tools a lot before November 2022 and they were

5:09

lame they just weren't very good so I we didn't find anything that we could really do with them and then November

5:16

2022 the first kind of public version of chadt came out we were on the waiting list for a long time we got in we got

5:23

super excited about what this thing can do and started reverse engineering it and figuring out what it can and cannot

5:29

do and immediately it was glaringly apparent that this tool can make a huge

5:36

difference uh for businesses and we've since then we've

5:41

gotten a lot of expertise in Ai and what I've learned is that people don't fully

5:48

understand the potential of what it can do for their businesses most people that even if they know about it which to me

5:53

it's very surprising that so many business owners and Business Leaders don't even know they may know it exists

5:59

but they've never even tried it they don't think it has anything to do with them or they're afraid of it or they just think about it as an afterthought

6:07

but those who do know about it mainly use it for writing for marketing for writing content and I think that's

6:15

probably about 30% of what we do with it maybe we do a lot of different things

6:20

with it that are not content creation and over the past year and it'll be year

6:26

and a half shortly we've been working with a lot of our clients who have also

6:32

been interested in jumping on this moving train and part of it was trying

6:38

to figure out how to implement AI in those businesses so that it will really

6:43

help us with things that we used to do before so we've always done business process implementation improving

6:49

business process and Automation and efficiency and and all of that and then

6:54

comes Ai and we're like constantly thinking of ways to use those tools to

7:00

take that step further or two steps further or 10 steps further and we've I

7:06

can give you some examples if that's something you're interested in yeah absolutely I was just going to ask if 30% is only and I think most people

7:14

think of tools like chat GPT as content creation right but it sounds like the

7:20

bulk of what you're helping companies with is other ways to to use AI can you

7:25

give us a couple examples I think the number one example on the biggest one is analysis just analysis and analyzing

7:33

data all kinds of data numbers and different data think about it as the

7:39

ability to take a bunch of databases with different data in them finding commonalities and be able to extract

7:45

insights from that data that would have taken a human many hours many days many weeks

7:52

many months to even be able to process all that data one example that we've done

8:00

sometime last year middle of last year so we worked with a large Medical Practice Group and they have several

8:05

offices we've used AI to analyze patient data so we stripped we did a few stages

8:10

we stripped the P the data off of anything identifying for HP compliance purposes and we basically used it as

8:17

statistical data we moved it to a different database and analyzed patient data to uncover all kinds of insight

8:23

about their patient behaviors about referrals about their marketing operation about what kind of procedures

8:31

patients tend to come in and do what procedures are more profitable lots and lots of both financial and patient

8:38

Behavior data at the end of the day we improveed this group's bottom line by 20% just by doing that we also helped

8:46

identify operational gaps that they've had and closed them we improved the patient journey in general by by doing a

8:53

implementing a lot of processes and we changed all the marketing practices based on the inside that we've gained by

9:00

implementing AI That's one example now are does are you using tools to do this

AI Tools and Their Applications

9:08

type of thing outside of chat GPT yes I know chat GPT has capabilities like that

9:13

typically needs to be apid driven um are there any tools you can share that

9:19

people should investigate uh for analysis so we mostly use the chatu BT

9:25

API but we also use other tools and I think one of the one of the advantages

9:30

that we have that that I've seen that makes us unique is that the the fact that we are experts in business process

9:36

and business process automation implementation Improvement that has nothing to do with AI right we also have

9:42

technical expertise and the ability to implement API and backend and write code

9:48

to do all kinds of things and the combination of these two things is pretty unique to the fact that yes we

9:54

know the tool we know the ability and we know the limitations of the tools we know how to design a business process

10:00

that will find an advantage and be able to to take advantage of of the tools but

10:05

also how to implement those things in the back end in the way that will take

10:11

full advantage of the tool and I'll give you an example in the example that we spoke about now this medical group was

10:17

using an EMR and electronic medical record system that was not meant for any

10:23

analysis for any anything any numbers any bi whatsoever they're meant to

10:28

manage records and so at the end of the day all of these systems are writing

10:33

their data into database so we had to break into the database in a way find

10:38

the database take the data out of that database strip it out of personal identifiable information put it in a

10:45

different database and then write code that would allow chpt API to access that

10:51

database so we can run certain queries on it that's an example of it's a multi-step solution but at the end of

10:56

the day we created a a dash board for the practice that they could literally

11:02

ask questions they could ask questions of Chad GPT on that data and we created

11:08

a dashboard with graphs and with whatever different things but that's an example so there are many tools involved

11:14

it's not just AI tools and I'll give you another example we work with a major content publisher and we've helped them

11:20

transform their process by integrating both like a graphic AI M journey in this

11:27

uh particular example as well as Chad GPT and some other AI tools there are

11:33

tools that are based on Chad GPT but are doing The Prompt Engineering in a different way that already exist like

11:39

koala riter there are different tools Claude which is a different Chader it's

11:45

a different llm that is doing things in a bit of a different way and have different cap capabilities and so we've

11:52

been playing with the different tools obviously there's Microsoft co-pilot which again is based on opena I Chad GPT

12:00

engine but is doing things a little bit differently there are a lot of graphic tools Adobe Firefly and some other tools

12:07

that we have created into that process and transform that process from a

12:13

company that was basically creating all the content manually now there's a lot

12:19

less people the content is creating created is we call it AI augment it's created by AI That's being instructed by

12:27

people but we get a lot of again a lot of insights from the way that the content is created that's another

12:34

example there could be many tools interesting the from a client facing experience as

AI in Data Analysis and Business Process Improvement

12:42

great as chat GPT is when people are accessing the data you talked a bit about that conversational nature have

12:49

you set your clients up I think you alluded to that a little bit of having them talk to the data or have it ask

12:55

questions to get what they want uh can you speak to that a little bit yeah so I think that's probably the the

13:03

superpower the secret source of using CH GPT accurately is knowing what it's

13:08

capable of and really knowing the limitations of it and especially in the last few months with the ability to

13:14

create Standalone gpts and now with their team version that allows you to use data that's private to you and is

13:21

not used to train their models those are big advancements that we're taking advantage of and we do several things

13:29

one we create those prompts behind the scenes so that the you don't do the

13:35

conversational right the dashboards that we created are based on prompts that are happening behind the scenes they're

13:41

pulling the data out and it's a good thing and a bad thing because once in a

13:46

while openai updates the model every couple of weeks and sometimes they break it and so when they break it they

13:53

sometimes make it stupid and it usually takes them a few days to to figure that

13:58

out out and make it smarter again but then you have sometimes a week where

14:03

okay nothing works the way it's supposed to work You' got to re-engineer all the procs and that happens that happened in

14:09

the last six months twice that it really broke our processes and it's part of the

14:15

risk that you have to take when you're automating and you're implementing these things into code is that you're going to have to to fix it when it breaks and the

14:25

second thing is that we need to experiment before we create a solution we take a while experimenting with what

14:33

kind of prompts what exact not just the prompts it's really it's all prompts but

14:38

it's really what kind of background you need to provide and teach the model

14:43

before you even start asking the questions and I think that's something that a lot of people fail to do and I

14:50

think that's probably the biggest tip I can give people is don't just say I need to write a letter write this for me do

14:56

this for me you've got to give it background you got to really have a conversation and you got to build in

15:02

stages on top of the background that you're giving it so we are feeding it

15:08

very often with a lot of background information for example uh this medical

15:13

practice right we are okay you are a business intelligence analyst in a

15:18

Medical Practice Group this is what the Medical Practice Group does this is the background of the group this is how long

15:24

they've existed this is how many patients they have this is where they operate in the country lots and lots of

15:29

information about and this is what we're trying to get out of it these are the goals what are the goals of this whole

15:35

thing that we're doing right you want we're trying to gain insights about patient behaviors and then another thing

15:41

that I would recommend that people do is you ask it you ask chbt to give you

15:46

advice on what questions to ask because those are some of the things that help

15:52

it almost like process the information that you've given it right now that I've given you all of this background

15:58

information about about this Medical Practice tell me how would I approach

16:03

trying to get this information out of this database of data what would you ask what else do you need from me so that

16:09

you can help me analyze this data and then you get the answerers and that helps you engineer The Prompt and then

16:15

after you do all this then you can create and now again like the token size has been increased so now we can

16:22

transmit more instructions and it's not a one- stage process it's not like you

16:27

push a button and it fits out the information it's a little bit like teaching a child how to ride a bike they

16:33

fall they get up again and they get up again and again until they get it and when they get it it's like this right

16:39

they get it and it's oh I get it it's like the light turned on so and this is the process that I see

16:47

when I work with business owners and try to teach them how to do this is the piece that they're missing and even though even my kids are

16:55

using Chad GPT but making it second nature to me now this is the first thing

17:01

I go to I have a question I'm thinking about something I'm contemplating something anything it's the first thing

17:07

I go to the first thing I do I love that the the idea of making businesses

AI in Business Decision Making

17:13

smarter using Ai and helping with the decision-making process and analyzing

17:19

data to make better decisions I jokingly say I couldn't be

17:25

running a business the way I am now a year ago because so much has changed and I'm

17:32

curious what for those that may already be running a business successfully for

17:38

years and are hesitant to get started what recommendations do you have I'll

17:44

tell you what first of all understanding that AI is not the goal it's a tool it's

Setting Business Goals with AI

17:50

a means to an end right that's not people come to me I want to implement Ai and I'm like

17:55

why I don't know I don't know why so figure out the why first you got to

18:02

Define your business objective to understand that in order to optimize the business process and certainly in order

18:08

to use AI as one of the tools to optimize that business process you need to understand what your goals are what

18:13

is it that you're trying to achieve and that's the number one thing people kind of skip that they go straight into okay let's put let's Implement AI Implement

18:20

AI in what AI is a lot of things AI is like saying oh let's use the cloud okay let's use the cloud for what

18:28

let's Implement a i for then understand that it can be a tool to augment your

18:33

human operations and not replace them like people are looking for magic Solutions it's not magic it's really

18:40

important to work with experts that will guide you through the whole process of

18:46

understanding your need need understanding the abilities of the tools and also their limitations because they

18:52

do have limitations from many different aspects whether it's security whether it's like

18:58

a lot of different things that you need to be aware of and then figure out a process that really aligns both the

19:04

business needs and the technology that's necessary to the strategy of your business and like I said it's not a goal

19:10

by itself and understanding the advantages and the

19:17

limitations is very important in this case got it what have been some of the

19:23

more interesting goals people have had that have come to you with uh things that they trying to achieve in their

19:29

business I had a prospect um that was working with us uh last year and they do

19:36

a lot of financial data analysis and they're trying to implement AI in in

19:43

that process so that they have less people creating those reports that's one example that's a

19:49

very not a simple goal to achieve but that's the very correct accurate goal that AI can

19:56

certainly achieve but they were concerned at the time about privacy and

20:03

like security data security of the information that they're that they're evaluating and so at the

20:09

time we couldn't really give them a good enough answer because the tools really didn't have it yeah there's CH GPT

20:17

Enterprise it's almost impossible to to get a seat on and now there's the T version that is possible to get a team a

20:23

cedar pretty easily and so that's something that can be done now that we

20:28

couldn't do four or five months ago so that's an example most of our

20:34

client most of them don't really have the right goal in mind when they start

20:39

thinking about this with us that are implanting that idea in their head and

20:44

with the exception I would say of one client that's been really spearheading this whole AI thing and she and I have

20:50

been geeking out over AI for the past year and a half in ridiculous ways

20:56

including in the middle of the night I found out that I can do this and this and this oh check this out look at this

21:01

screenshot and yeah very geeking out and in a way I feel like working with chpt

21:06

is like having a positive version of the Borg at the edge of your fingertips

21:12

right you have this Collective huge mind of millions and

21:19

millions of people and millions and millions of seals and all you have to do

21:24

is know how to ask the questions and I think in our business knowing how to ask the questions is our sort of superpower

21:30

in the very beginning that's what we do really well and so it just extend into

21:36

Ai and that's why we are we've been successful with it I think it lends itself to the years

21:43

of experience you have in process Automation and really understanding bi systems because without having the right

21:49

questions you can be misguided what makes a good business goal when it comes

21:54

to AI I don't know that there's a good business goal when it comes to AI I think your regular business goals that

22:00

your day that's what I'm saying like your usual day-to-day business goal most of them can be improved or made more

22:09

efficient with AI anything from okay marketing and I'll give you an example from our own world we have been we're

22:17

operations and NH HR technology people and we're not very good in sales and marketing so trying to Fig figure out a

22:24

way to grow is that we're analyzing our funnel and trying to understand what works better what doesn't work why it

22:30

doesn't work one of the ideas that came up is let's send a survey to people that we've sent proposals to that have not

22:36

signed up with us and then I'm like yeah marketing surveys are a serious business this is not our area of expertise let's

22:43

figure out how to design a marketing survey that will really give us what we want to hear and how do we do that we

22:50

ask CHP so the goal is the same goal it had nothing to do with AI but the

22:56

implementation of the goal had everything to do with AI because Chad GPT gave us a lot of ideas and a lot of

23:03

things that we haven't even thought about on how to phas the question so it is really giving us what we want there

23:08

are multiple ways to phrase a question you're not always going to answer what you actually want to know and so that's

23:15

one example another example is that we're working we're building together with with client a

23:21

nonprofit Association for web Publishers that will help represent web Publishers

23:26

small Publishers and their interest in with big Tech and with different

23:31

different things and we are working on the bylaws of the organization and you

23:37

have to think about all the things that could go wrong right when you put together a board all the things that could go wrong all the things that you

Utilizing AI for Board Management

23:43

need to think about and Chad GPT is a perfect perfect companion for that

23:48

because you ask tell me all the things that could go wrong when I'm hiring board members help me put together a

23:55

conflict of interest policy what should they be allowed to do and what should they not be allowed to do and again it

24:01

doesn't do the work for you but it gives you a ton of ideas and a ton of information that you otherwise would

24:07

have taken hours and hours to research got it this is great I'm really enjoying

24:13

this conversations tell me like me yeah total geek and that whole thing you were

24:19

talking about with like your your little friend group there where you're constantly like messaging people with have you seen this have you checked this

24:24

out I have the same thing when it comes to the the way that you're working with

The Pros and Cons of AI in Business

24:33

clients now have you found that AI is creating more work in certain cases when

24:39

it comes to either reviewing things or I guess what I'm asking is it always

24:46

a Time Savings or where do people need to be careful with AI in creating unnecessary work certainly sometimes

24:53

it's not worth it right because of what I said before that it's not a magic

24:59

solution you go in you ask a question you get the answer it takes work it takes time my interactions with Chang PT

25:04

even about smaller things they take a while they take a long time sometimes it's not worth it I can write the email

25:10

myself I can write the document myself sometimes I don't need CH GPT to do it

25:15

because it will take me more time to teach it everything that I have in my head that it needs to know to have perspective in order to create what I

25:21

needed to create so yes definitely sometimes it's not worth it in other cases I would say the process itself of

25:30

figuring it out and implementing it and creating the processes and creating the prompt and writing the code and all that

25:37

takes time that's again not a magic solution and we'd always have to weigh

25:43

whether it's worth it and in many cases not just whether it's worth it but also

25:49

how what's the lifespan of the solution because you put in all this in a solution you're automating you're

25:54

creating code and then they make a change to the model and and they break it and then you have to maintain it so

26:01

sometimes it's definitely worth it because the amount of time that it saves is very significant and the amount of

26:06

money that it saves is significant or when it does things that a human simply can't do and that's common but sometimes

26:14

it's just really not worth it and we decide to not Implement like a software

26:20

solution but we decide to teach people how to use it manually because the constant changes make it so that it

26:26

doesn't make sense so it really depends on what it is you're trying to do and I tell you that in the case of the this

26:32

publisher I was talking about we decided to not automate the image Creation with

26:38

mid Journey we decided to keep those manual because the work that it takes to

26:43

automate them and the changes to the model and to the language constantly

26:49

they're too rapid they're too quick and so it doesn't make any sense to sit there for a month and write a piece of

26:54

software that every time they're going to release a new version It's not it's going to be absolete completely agree

27:00

with you in that there are certain cases where it doesn't make sense and you get

27:05

there through experience I remember when I was running a PR firm we got into the habit of doing these custom proposals

27:11

but each time you do something long you got to read through the whole thing right you learn that after you realize

27:17

why is this taking longer than it used to yeah right there's a reason templates exist they're designed to save us time

27:25

there are a couple things I want to touch on one is the future what in your mind is

The Future of AI: Privacy and Training

27:32

the lwh hanging fruit for Solutions around AI I guess those

27:38

are two questions let's talk about the low hanging fruit first and then we can touch on the future I think the

27:44

real dramatic change is going to come when Chad GPT Solutions can be not

27:52

really self-hosted because you can't really self-host this computing power but really be segreg ated and completely

27:59

private from the rest of the world that hasn't happened yet and although your data isn't being used to train the model

28:06

still not 100% your data and I think that's still a major limit limitation of

28:12

all of those llms is that you don't have a way to really make it completely yours in a legal perspective so I think that's

28:19

one thing that's really going to make a huge change is when that happens and I hope that happens this year uh because

28:25

it's the whole thing's going to blow up when that happens when you can really control that your data remains yours and

28:32

remains completely confidential and then you can expose the model to things like patient data you didn't have to do all

28:38

the things that we did and move it from one database to another like all of that in order to protect uh

28:44

confidentiality the other thing is make it in a way that you can really

28:51

train the model on your environment in a way that you don't have to keep repeating yourself they did that with a

28:57

gpts but it's very small the amount of information it retains is very small

29:02

it's not enough and when I'm actually able to host a llm model on my own environment Cloud environment obviously

29:09

and train it on my entire library of data whatever it is where that data is

29:14

going to be retained and it's going to be able to pull it and it will have the

29:20

Computing ability to really scan it in real time all the time I think that's

29:26

going to be a huge major change to the whole world but I think that's going to take some time just because of Hardware

29:32

limitation very interesting there was a a tool that I came across recently I

29:37

haven't explored it too much Julius AI

29:42

all data and Analysis something that I definitely would love to dig into I'm

29:49

not sure you mentioned you haven't something for us both both to dig into after this and maybe on our next

29:54

interview we'll we'll be able to chat a bit more about that one but there's always tools coming out how do you and

Keeping Up with the Rapid Evolution of AI Tools

30:01

your team keep up I know it's challenging I often love doing podcasts

30:06

where I get from the horse's mouth so to speak of oh there's you got to try this or check this out but I know being on

30:14

LinkedIn there's 50 prompts for this and 100 prompts for that and 10,000 it's I

30:20

just need to relax a little bit and have some perspective uh where do you how do

30:26

you navigate the influx of uh new yeah

30:31

so first of all we have a group within our company that is just informing

30:36

everybody on new tools that they've come across whether it's on LinkedIn whether it's heard it on the radio whether it's

30:42

in a podcast whether it's whatever so we suggest them then we take a look at them really quickly to see what they're

30:47

supposed to do and we decide whether it's worth testing them out or not based on whether we think it's going to be

30:53

relevant to our business model there are certain kind of buckets of tools that we have been testing extensively I'll give

31:01

you an example not takers for Zoom for Google meet for whatever I think we're using four different ones and they all

The Role of AI in Note Taking

31:08

have their advantages and disadvantages and we are literally using all of them I

31:15

I laugh now because often it seems like at meetings we're outnumbered by note takers right they all just flood in and

31:22

our our screens are so small because it's full of note takers but what are the different note that you're

31:28

experimenting testing recently Zoom added their own AI summary feature and

31:33

the bot I forgot what it's called that you can ask questions during a call that thing is a game changer Game Changer

31:39

turn it on because if you doze off for a second and your mind wonders and somebody says your name you're like oh

31:45

what did they just say that's super cool don't tell anybody I said that I won't

31:51

in addition to that we are using two tools one is called fathom Ai and the other one is called read AI they just

31:58

have different strength in in really pulling out insights from conversations and I

32:06

find myself using all of them the summaries the Highlight the action items

32:12

all of this I'm using both of these I'm definitely using Zoom we are also using something called metav view or HR and

32:19

interviews where it really does a very good job in capturing an an interview because we interview a lot we do a lot

32:24

of HR we hire and we recruit for a lot of clients so we use that one and they're all being used by different

32:31

members of the team at different times so that's no tiers we've experimented with I don't know hundreds we have a

32:37

Monday board on our um Monday instance where with all the different tools that we've experimented with and whether

32:44

people who's who've tried them decided that they're worth it or not worth it those that are worth it we register we

32:51

tried them out we see so we tried all kinds of tools for video production and

32:57

played around with them obviously all the image generation tools logo and design tools every time something comes

33:04

out we we try it if it makes sense for our world our business world and I know

33:10

that a lot of our clients are using tools that are helping with troubleshooting code and writing code

33:16

and documenting code and things like that I don't write code but so there are different tools I think the qu the

The Importance of Continual Learning in AI

33:23

answer to your question is that we just like it we really do enjoy it we've implemented on our website over the last

33:30

several months five or six different AI tools and some of them lasted a few months and disappeared okay we go on to

33:36

the next one but we like to experiment and try them out I think that's a great point in that

33:42

there's some people that have always had an appetite for Learning and AI is just

33:48

like the if you like chocolate it's the most delicious piece of chocolate that you can keep going back to and I'm not

33:57

surprised that for yourself and your team that you do geek out on this stuff

34:03

and when it first came out it's almost like you can't get enough even now because it's

34:09

constantly I don't think there's ever been something that you keep meing to refresh what you know or rethink things

34:17

and there's probably as far as safe bets some companies

34:24

that I think as far as Maybe correct me if I'm wrong but chat GPT is a good one

34:30

to learn about as it's probably not going away Absolut that's not going anywhere are there any others that you

34:37

would say are good places to invest time

34:43

that you're predicting are going to uh still be around I think understanding

34:48

the buckets of the different tools like the llms the image Generations the the audio and video generation like

34:54

understanding the tools and who the major players are is the the first kind of Step because you want to know who the major players are on a financial basis

35:01

as well who's backing who Microsoft is backing open Al Al you got to know these relationship that's just like business

35:07

curiosity and if you're following business world you're going to know these things that's one thing and the

35:12

second thing is once you understand the bucket keep on trying because I'll give you an example in the very beginning I

35:18

was like all over Bard and Bing and this and that and the other and it was garbage so I left it alone but I I have

35:25

a calendar reminder every quarter to go back and play again with them because

35:30

maybe something changed and I can definitely see how for example being

35:35

able to use AI within the Google Suite which is already implemented but it's

35:42

still still not entirely there but being able to use it within my Google Suite on

35:47

my documents that's going to be amazing but right now I still go back into Chad GPT you know if I had that within rle

35:54

Suite I would be a very happy camper but it's not there yet so reminding myself to keep trying those tools that maybe in

36:01

the beginning I said okay it's really not there is a good example anthropic Claud is a good example because it's a

36:07

very different I know it sounds ridiculous but it has a very different personality and so sometimes that

36:13

personality comes in handy for things where Chad GPT you'd have to teach it to have that personality and Claude already

36:20

has that personality and so reminding myself to go back and look again at

36:26

tools that I already decided not to go with is something that I do it does seem

36:32

like a crazy bandwidth thing as a solo entrepreneur to try to navigate all of

36:39

these different tools and to your point it is a great idea to revisit things I

36:46

personally haven't experimented with Claude very much just a little bit scratching the surface and but I keep

36:54

talking to people and it comes up again and again I think that's also a good indicator is that if a tool keeps coming

37:00

up in your ecosystem then third time you hear it maybe sign up for a a trial or

37:07

get your feet wet but obviously chat GPT comes up over and over so everyone's

37:13

been there to your point about tool integration I was using I like notion

37:18

for organization and I was using their notion Ai and it's powered by chat GPT

37:27

and I love the the organization of being able to not have to copy paste but

37:34

there's a familiarity now I think a lot of people have with how chat GPT works

37:40

and the expected output that keeps bringing me back at least and makes it

37:48

not like I'm ready to just yeah implement it I like having a history

37:54

there of everything that I've been working on although I still don't maybe

38:00

I'm missing it I was using a browser extension allowed me to search my

38:06

threads but having a search in chat GPT seems like a big oversight and maybe

38:13

there's a reason for that yeah I would also like a copy the thing at the top not just at

38:21

the bottom of the the results so that you could do it in two places copying is

38:28

usually something that if I'm starting there's definitely some UI things that I wouldn't say the de UI is genius UI not

38:35

at all but they've added some where what's happening is that I used to use a lot of extensions that now are built

38:42

into Chad GPT such as sharing a conversation or or summarizing things

38:47

like that that I used to have all the I used to have 6,000 browser extensions that would do all these things I don't need them but yeah that those things are

38:55

going to be helpful and the ability to pin or reorder your conversations on on the left would be very helpful and but I

39:03

also found that in older conversations this is also new by the way in older

39:08

conversations you're able to switch the model you weren't able to do that before that's new that's true and in older

39:15

conversations it forgot the context so you'd have to remind it to go back to

39:20

the context so like there not smooth sailing yet I'm finding little things

39:27

and bugs on a daily basis but then again if if you again it goes to my point

39:33

about understanding the tool and the limitations and what it can and can't do and once you're able to do that I think

39:39

that makes a big difference we stress test every new model that comes out every new version that comes out we

39:45

stress test we do whenever we go into any new llm or even something that's

39:50

built on top of the API but has a different UI because they're doing

39:55

something we stress tested we asked all the inappropriate questions we try to see what it will censor what it will not

40:03

censor we try to check whether it has accurate information we ask for references so there's a lot that you can

40:10

should be doing before you feel comfortable and even then even when you

40:16

do feel comfortable I think you have to remind yourself to check it again and again because it changes good advice we

40:24

were talking about note takers earlier and I had a conversation with someone and a notetaker came up so there's

40:30

here's one more for your your team to check out it's called cibil AI or Sy YB

40:36

I L I believe and it's a notetaker but its primary focus is for sales teams so

40:44

it's got one kind of mind reader type functionality where it somehow I haven't

40:52

I'm still new to it but it's supposed to predict pain it will autogenerate a an

40:58

email that's drafted ready to go based on that and I it's all these micro steps but it's really focus on the sales

41:05

process with tight Integrations into crms and that's one to take a look at

41:12

that's very cool from a organizational perspective have you heard of Tas a yes

41:19

this one's come up a few times and for our listeners always trying to give some value as it relates to tools

41:26

this one's supposed to have some incredible functionality as far as multitasking of gpts to get things

41:33

moving simultaneously where often I don't know about you but sometimes you're like waiting for chat GPT to to

41:40

do its thing right lately it's been so slow it's been driving me crazy yeah so

41:45

as great as chat GPT 4 is sometimes the speed of chat GPT 3.5 yeah does what it needs to do yep

41:54

that's correct this has been amazing you are are great I love learning about this stuff and I've had such a wonderful time

42:00

chatting tell me where people can get a hold of you your website what's the best

42:05

place to contact you so our website is shac consulting.com sbac consulting.com I am a d axman on

42:15

LinkedIn and that's the only the only advantage of having a weird unique and

42:20

foreign sounding name is that you Google my name you'll find me right away and we're everywhere we're all over the

42:26

internet and so it's very easy to find us and if anybody needs help with AI or otherwise business process or anything

42:33

we'd be happy to talk to them we we don't we're very picky with the clients that we take because working with a

42:40

fractional executive is really a partnership almost like a marriage it's

42:46

a real partnership and so some people are just not really ready for that and people that are ready for that will reap

42:51

all the benefits because it's a really great model I've been doing this for 15 years and it's a lot of fun amazing

42:59

before we wrap up I am curious what you're reading are there any books uh

43:04

that you maybe go back to or would like to share I'm reading all kinds of things

43:11

I'm mostly reading magazines and whatever I a lot of my business colleagues and people that I know in

43:17

business and especially in my age group are now writing books so a lot of the books that I'm reading a friend of mine

43:23

wrote a book about fractional leadership so that's a great book his name is Ben wolp and so I'm reading

43:29

a lot of these kinds of books and I'm also finding that a lot of books that

43:35

are out there now about business have been written by Chad P yours truly and

Identifying AI-Generated Content

43:42

which means that if I know that it means that some of them haven't been doing a very good job because if you do a really

43:49

good job with using Chad GPT no one should be able to realize that you wrote this book with Chad GPT what are the

43:57

flags for identifying something's chat GPT I don't know if I can make a list of

44:03

them but I would start with I hope this email finds you well no and there there

44:09

are just some telltale signs in the language that to me show that you

44:15

haven't trained the model to be you you know what I write I hope this email

44:21

finds you well so maybe I'm a little I got that in me but I I do add a

44:27

yeah CH got it from somewhere yeah but there are some telltale signs that are

44:32

in the language um I would as advice to people that are using chpt to write is

44:38

that read what it writes and if there are certain words which always are that are not you that you're reading this and

44:45

you're like I would never use this word don't use the word and so I can't there

44:50

are a lot of people that I and I'll give you an example I had an employee um at a client and she wasn't she was an um

44:58

native English speaker and she wrote me this really long email that was an absolute perfect English in grammar and

45:04

I'm like she didn't write this and it makes you like it makes you feel like I know this

45:12

person I've been talking to her for a long time and I've been emailing back and forth with her for a long time all

45:17

of the sudden this email is completely not her right I would tell people like if it

Maintaining Your Voice in AI-Assisted Writing

45:23

doesn't sound like you even if it sounds good or better don't use it because then

45:30

you lose your personality and that for me and personally that's something I refuse to do if there is a word that I

45:36

would never use then I tell Chad GPT I have a list of words don't ever use these words when I'm asking you to write

45:41

something as me that's never them that's great advice is if you are using it for

45:47

writing maintain your own voice so that you come through and

45:53

aren't in a way take taking people out of whatever they're there for and if

45:59

they're trying to learn something from you they shouldn't need to ask that okay

46:04

this isn't her or him or whatever so it's also about the attitude right like

46:09

I admit I'm I used to be Israeli and I have an attitude and if Chad GPT strips

46:15

me of my attitude you read something that's super nice and cuddly that's just

46:21

not me so why do I want to represent myself as something that I am just not I

46:27

don't see how it's going to get me anywhere and I also think that it's been salting to the other side that's

46:32

receiving whatever it is because it's not me so I think maintaining your voice and learning the tool enough to know how

46:39

to maintain your voice and I will give you an example when I write using chpt I give it feedback all the time I would

46:45

never use this word I would never write this this is too nice for me this is too salesy for me this is too touchy feelish

46:51

or me this is to give it a lot of feedback all the time amazing this has been an absolute pleasure thank you so

46:58

much for joining me and I look forward to doing this again sometime