Use automation to build optimal flow states with Shawn David

AI-Generated Summary

Learn about human-first engineering in this in-depth interview where Shawn David discusses strategies to organize work and to break down tasks into smaller digestible sessions. He explained the most effective ways for teams to discover the work that should be automated the most by focusing on the desires and strengths of individuals, to remove work that is menial especially to them. We explored the impacts on company culture that menial tasks can cause and discussed some of the pitfalls with new technologies as well as lessons to keep in mind while reviewing new tools and technologies.


AI-Generated Transcript

Use automation to build optimal flow states with Shawn David – YouTube


(00:00) menial tasks by definition don’t need a human to be done so if your job to be done is high level cognitive thought and the tasks that you’re continually assigned are jobs that do not need high level cognitive ability that is an immediate red flag for everyone about the quote unquote culture of the company hi my name is Romo Santiago and

(00:33) I’m the founder of experiment Nation today learn about human first Engineering in this in-depth interview where Shawn David discusses strategies to organize work and to break down tasks into smaller digestible sessions understand the most effective ways for teams to discover the work that should be automated and how all this impacts

(00:50) company culture enjoy the episode Sean David welcome to the experiment Nation podcasts our listeners might know you for talking about processes and Automation and how to make work easier uh would you like to introduce yourself a little bit more to our audience uh sure yeah hey I’m Sean David owner of

(01:09) automate to win and uh I Pioneer the process that I like to call human first engineer and U basically what I do is I go in and I try to identify bottlenecks and busy work in people’s lives and then remove them with automation so so that’s what I do awesome thank you for that intro um so right off the bat I’d like to get

(01:34) into our first question and try to understand if there’s any strategies you’d recommend uh for people when it comes to organizing their work and breaking down tasks into smaller and manageable bits so that they can be effective and efficient in delivering their work yeah sure so the the first thing that I like to do is make sure

(01:52) that we identify what we consider like a session or a seat so my personal opinion the way I look at this is I don’t think anything is really able to be done in less than like a 15minute block so even if you’re going to respond to an email it’s going to take you a certain amount of time to as pedantic as this is sit in

(02:10) your chair start your computer open the email browser open Outlook go through there’s steps that have to happen there right so even if you say checking email only takes me uh two minutes let’s say when I log in for the day sure that may be true but the thing is you’re not measuring all the periphery things that

(02:34) happen around that so when we look at processes what we have to do is we actually have to just reduce them to their most idiotic and insane steps because we’re humans and we tend to forget step Z one two three four exist because we only kind of click on at step five so for instance like I you try to explain

(03:02) somebody to check their email you say okay I sent you an email check it you’re operating under an assumption that they understand what zero to five or six is they understand what an email is like it’s funny but you know they understand how to how to log into the computer they understand all of these things by simply

(03:23) saying hey check your email so what I like to do is break down things into 15minute blocks and say what are you doing in this session time at 15 minutes and then through that you can list out every single thing that you’ve done and you can kind of break those into quadrants of an hour if you do that for

(03:42) an entire week you can start to recognize patterns right you can start to see this and you can start to say I spend this 15minute block here and this 50- minute block here and this 15- minute block here you can start removing combining altering looking at all of those different blocks none of this has

(04:01) anything to do with automation or AI or anything this is wrote stuff where you do this yourself and this can’t be done by a consultant it can’t be done by someone else it can’t be done by AI it can’t be done by anything but you you have to go through and say honestly this is what I’ve done for the last 15 minutes 30 minutes one hour

(04:22) whatever it is that you feel comfortable with and then you go through that and you prioritize the things that you’re doing to understand what the Step Zero to step five that you’re forgetting and then that’s how you you can prioritize what to look for because the problem is people want to look for these processes

(04:41) that take a really long time and they want to eat the elephant and they it gets too complicated and they want to be like oh man I need to automate my email okay what does that mean what does that what does that even mean to do I I need okay do you want to automate responding to emails do you want to automate

(05:01) categorization do you not want to get spam do you do you want to block time so that way you can’t check your email for a certain time because you know it’s like a path lob trigger for you like what does it mean so you have to go through and identify this and how you do that is you you take you Journal right

(05:18) like if you it’s just like anything else right like if you want to optimize your weight loss Journey what do you do you you write down what you ate for day and at the end of the week you look back and you’re like oh man during this time and this time I really I really have not an optimized process I need to look at that

(05:37) but until you measure until you break that down into individual things you’ve eaten then you’re never going to be able to actually do anything with that data right so how do you do that break it down into 15 30 60 Minutes whatever you feel comfortable with and then journal what you’ve did and if you’re not

(05:56) comfortable with that that means means that you’re not comfortable with optimizing your time and there’s a reason for that and that’s fine but if you’re looking for like automation this is Step Zero this is Step this is you have to do this um and if you’re looking at like automation as like a something

(06:12) to solve for some problem it may not be the it may not be the solution and the problem may not be you have too many things to do right so identifying the steps that you have the stuff that you have to do in those blocks of time that’s the that’s the the way to go that’s the only really way to start all of this you

(06:32) know you gave two examples with opposite time durations one is very short one is very very long and I guess the direction we’re trying to go with this identification process is to help us at the end of the day build a workflow daily that allows us to get into a a working flow and maximize our energy

(06:51) that we spend towards work right yep yeah you want to ident it’s just there’s no real right the wrong way because if you’re doing like let’s say you’re doing video editing right or something that’s like super highly creative like graphic design or graphic editing or video Whatever all this stuff right none of

(07:10) those all of those things take a really long time and every session that you have is going to be significantly longer than 15 minutes like you can’t you just can’t like let’s say you’re a twitch streamer right and you want to optimize your processes like 15 minutes is nothing to you like you you need to look

(07:27) at the processes that you re regularly do uh within your systems that are a chunk of time because you you have to and and get this correct though once you have identified those things that take the chunk of time you have to break those down at individual steps like the second microsc steps but you have to identify

(07:51) the processes of the things you do and if you’re in a regular business and if you’re in a marketing or advertising or a development agency something like that you probably have rot quote unquote busy work tasks that have to happen that you can identify through pattern recognition and it no one’s going to tell you how

(08:10) long that can be so I I don’t think it should be longer than one hour because that way you can easily recognize okay every hour you should be standing up anyways so you should look at it that way you every time you take an hour you optimize the system just to for your brain um so if you do that like I I

(08:26) would never recommend going more than an hour but look at everything that you’re doing and and maybe you have a 4-Hour session of deep work but that would be four separate hours of these things and what did you do during those hours and it’s not necessarily short time or long time because there is no short or long

(08:47) depending upon what you’re doing so if like I said if you’re doing video editing a a short amount of time may be one hour but if you’re doing email marketing a short amount of time may be five minutes so but that five minutes is never five minutes it’s always periphery so what I what I recommend is going

(09:05) through and identifying and writing everything down and prioritizing it for the reason of understanding what it is that you’re actually doing to build the system to build the workflow saying okay I understand maybe you don’t even realize that you check emails that between 10:30 and 11 every single day

(09:22) maybe you don’t even realize that but then you look at the pattern you’re like oh I stop what I’m doing right before lunch and I spend 30 minutes every single day just going through my email and maybe you know you just don’t even recognize that because it’s just what you do so what you’re trying to do

(09:38) is you’re trying to break the pattern by recognizing the pattern and then once you do that you can start optimizing because you realize there’s a pattern so the time frame only is relevant to the person doing it it’s not necessarily like a hard and fast rule if that interesting um it’s on the topic of time

(09:55) and the really relativity of the task duration based on what you do how helpful would you say is time tracking when you’re working with teams especially knowing since that a lot of people are actually uncomfortable with tracking their time feeling that they’re being you know watched and kind of supervised how helpful is is it for you

(10:16) to have time stamps and understand the relative importance of every task in a in a work week this depends on the on on the systems and you know this is why it’s human first and this is why there’s no right answer ever time tracking is amazing if you Comm monetize time okay if you’re in a system where

(10:37) it’s transactional and you sell time uh lawyers doc like not really doctors lawyers mechanics um anyone that does flat rate like if you’re in the aviation industry if you’re in the automotive industry if you’re in the legal profession um if you or in anything that has a direct Roi to commoditization of

(11:03) time um everyone that is involved in that signed up for that that is the show that they bought tickets for so to not have it in that in that system is uh disservice to the users of the system and they want it why do they want it because they are paid strictly for their time it’s a transactional thing like for

(11:27) instance if the listeners don’t understand what flat rated is it’s really simple like when you go to the mechanic and they tell you that it’s going to be uh let’s say 10 hours to replace your engine in your car that mechanic gets paid 10 hours to do that job if that mechanic does it in two hours that mechanic gets paid 10

(11:48) hours dope if that mechanic gets does it in 20 hours that mechanic gets paid 10 hours okay the reason that you want to track time very very clearly in that situation is because if I get paid 10 hours to do a job I can do in two hours that means that I can do four of those every day and I can get paid 40 hours

(12:11) every single day so you bet your bottom that I want time tracking I want the specific time that I’ve put in that I could sell because I’m getting paid for this so if I took two hours I’m a track for two hours and they get they sell 10 hours and now I’m upside 8 hours I can just go home for the day I made 10 hours

(12:31) or I can continue okay legal profession of the same way a legal profession uh they work on a retainer basis which means you give them a grip of money and then they work off of that money so you say here’s $5,000 or here’s $10,000 and they’re like all right cool that’s just for us to exist and you get for in that

(12:50) you know you get this guy and he’s the main legal guy and he’s $150 an hour and then here’s the Searcher guy and he’s a $100 an hour and here’s this guy okay and they all get paid by the hour to a certain amount above the retainer so if the retainer says that you get five hours of research every month and the

(13:09) researcher finds a way to do 10 hours that means that that researcher gets paid a bonus for the time above that they tracked right so commoditization of that time and the transactional nature of all of those systems means everyone wants it everybody and if you don’t have it you’re very upset that you don’t have it

(13:35) because you think you’re leaving money on the table it’s like I need this like this is what I’m doing like these are my receipts like I need the balance sheet at the end of the day for me to be able to give this to you so I can make the money for the system okay on the converse on the opposite side of that if

(13:52) you do not have a trans a transactional nature which between you and your employer your employer and you if it’s not transactional and commoditization of your time and they’re renting your time time tracking at that point is a burden because you are judged on your overall contribution of your time rented

(14:15) to them in relation to the to the overall cost of this balance sheet thing called Labor you are a to them okay and you are going to have sort of an issue there if you don’t understand that you’re just renting time and if you put like for instance you put your selfworth into whatever it is that you’re building and

(14:41) not understanding that what you’re building isn’t relevant to the person that is um renting your time the time is the thing so there’s a conflation of apples and oranges of what’s being sold and what’s being bought so time tracking if the time tracking is a transactional commodity is perfect and un needed and

(14:59) is an amazing thing but if it’s like a salary thing where you’re not really judged on your output but your input then you have to just kind of find stuff to fill the time because it’s not transactional that time has been rented regardless of if you have something to do or not so this is like why the the top- down management

(15:23) style requires you to um be in the office why they want everyone to be in the office and why middle managers want everyone to be there is because the middle manager in the MBA understands that they’re renting your time and that you are not a commodity your time in aggregate is the thing being rented and

(15:45) the output is the aggregate of all of that rental time and the way to be efficient in that rental time is to be uh fill it and it doesn’t really matter um the output of it it’s the um the Delta of available time to fill time and you’re measured as a middle manager by that so you have nine people that you

(16:10) have underneath you and their uh menial output is 60% and your kpi is to get their medial output to 70% at that point you you have to either decide do I go and have like hard conversations with people or do I just fill their plate with busy work and increase that number so time tracking it just depends

(16:39) on the system you know and what you’re doing that’s why it’s humans first right the tools are going to depend on this specific situation right so if we go back to teams where we looking to optimize their work what are some effective ways they can use to identify things that could be autom take my automation

(17:00) quiz so I uh I just spent 200 hours building this thing because this is like honestly the the the most um frequent I guess would be but the thing that I am asked most by humans is what can I automate what can I automate they learn my my company’s named automate to win they they run through me through Linkedin they you know they hear

(17:22) me speak at something and they hear me talk about you know removing busy work to help you human first engine Ing and they’re like hey great and everyone’s happy but then the next inevitable question is well what can I automate and the answer to that is you didn’t listen to this whole first part because it’s

(17:37) human first like you can automate whatever you can automate like I can’t help you with that because your processes are your processes and they and they and they got you to wherever it is that you’re at right now so to reduce those as to say they’re not working is not is a trap it’s not the right way

(17:57) what what you have to look at is say all right where I’m at right now is not necessarily where I want to be and not even where I could be because you don’t know where you could be it has to be where I want to be so what I did is I built this nine question uh quiz it’s automat automation. quiz um automation quiz what this will

(18:18) do is it’ll ask you nine questions and it’ll ask you how frequently you do one of these nine things so like how frequently do you copy paste between one program and another how frequently do you report into multiple tools how frequently do you create assets how fre how frequently do you organize or categorize your email how frequently um

(18:39) do you you know synchronize or create calendar hookups stuff like that right and it’s hourly daily weekly monthly never and what that did is I built an algorithm that goes and it comes back and it identifies tasks that you can look for and says these are the tasks that you should definitely look for

(18:59) based upon what you’re doing so like if you’re a marketing person and and you’re doing a bunch of copy paste what I would say to do is uh first of all enable sticky keys and if you don’t know what this is if you’re on a PC you hit shift five times and if you’re on a Mac you go into systems uh settings keyboard

(19:18) accessibility and set on sticky keys what is sticky keys it’s a it’s just a it’s an accessibility tool so let’s say you have one finger and you want to type how do you copy paste cuz you you can’t like use your nose or your finger like how do you do so this is their way around this and uh sticky keys is literally what it sounds

(19:41) like hit shift five times or turn it on the Mac and it allows you to do control and then it’ll hit it you’ll hear a noise and that noise will tell you okay it’s now listening for the second so youit control beep then C and then now you’ve copied so then now you hit control B and then you can hit V again

(20:01) and now instead of holding control and hitting V with you fingers you can use the accessibility keyboard for um copy paste right why does this matter and why do we use it because we set up a Pavlov trigger and everyone knows what Pavlov is the Bell Rings dogs Elevate no problem this trigger is okay I hit

(20:21) control to copy write down what I just did so in the automation quiz there’s literally nine worksheets that you print out and it’s annoying and you have to get a printer but there’s a reason the reason you do this is because when you write something it’s easier to remember and it breaks the pattern and it helps you

(20:42) understand what you’re doing so you go in and you hit control C and you’re likeo oh okay I’m copying something what did I okay I’m in Excel and I’m going into Gmail and done all right five minutes later going Boop geez okay and except so then you go through and you literally have a list at the end of the

(21:04) day of 50 or 60 times that you copy and paste it and you’re just like why did I do all that and then if you do this for a week you can see 50 60 70 80 times a day that you’re doing this and the thing that people don’t realize and the thing that’s that’s so critical about this is the time that’s built in around that the

(21:27) code switching that you have to do you don’t even think about because we’re humans and we just dismiss all of these things but you have a minute to a minute and a half of just I’m in a UI of Microsoft Excel and there’s a ribbon and there’s you know cells and there’s like I’m in a spreadsheet and now I have to

(21:48) go copy data from a website into a Microsoft Excel that’s something that you probably have to do on a regular basis right that does doesn’t seem like it’s a big deal it seem like we just process this like it’s no big deal but let build an AI agent to do that like build an AI agent to do that build an AI agent to go into a website

(22:10) grab a specific contextual thing on that website based upon a search from work that’s already being done format that data into a shape that makes sense then paste it relevantly into the spread sheet contextually for whatever you’re doing format it to remove links and remove strip the HTML all of those

(22:32) things right now go just pause this and go build a GPT for that and come back to now and then tell me how how easy that is and you’re doing that 80 times a day you like the mental load and cognitive buildup that that is you just have a certain amount of decisions that you can just make there’s a certain amount of

(22:54) mental fatigue that you can handle so if you find yourself doing this stuff this is like this is the problem you you don’t look at I need to automate the copy paste you look at I need to automate the thing that you’re copy pasting into like the end result of the thing and that is never ever ever going

(23:17) to happen because you’re just never going to get started you’re going to look at this and you just be like I don’t even this is too I don’t even know what I do I what I do and because you’re because you’re starting at step five and you’re start you’re telling the agent or the the automation to start at step five

(23:30) and it doesn’t even understand what step one is so it’s like I don’t know how to do this you know so that’s the that’s the Trap there is we forget we forget how complicated the stuff is that we’re doing because it just happens inside of our subconscious because we just inherently do it but if you’re code

(23:50) switching between different uis that’s cognitive load if you’re having meetings it between any other data that’s that is complete 1,000% will burn you and you don’t understand this but what’ll happen is you’ll be in investigative data analysis mode and that inquires zero social interactions that inquires zero

(24:12) um interactivity or active listening or any any kind of cognitive ability that requires you to be like present that just requires you to be a robot you can put music on it and then you can be surprised at the end of it how you just got to the the end result because you just get into autopilot if you’re broke up of that

(24:33) right usually takes about an hour to get into that mode if you get broke by that because you have a meeting or because somebody distracted you or because whatever getting back into that is never gonna happen you’re you you may get into a similar state that allows you to like um interact with with the data and and

(24:54) the result would be the same but the process that you were in of that flow state it has to be a flow so you you have to understand the difference between cognitive and non-cognitive system one system to all of these things to start identifying the patterns of what can I break out of right so how what do you do here take

(25:20) the automation quiz get these nine different things say okay these are the different places and different ways that that I am um basically patterning my time and then if you have let’s say Tuesdays are four hours that you understand that you just generally aren’t interrupted not really doing copy pasting you’re not really do then you

(25:44) look at that and you say um I got four hours that I’m gonna block a meeting with myself that no one can book that I’m going to use tools that don’t allow me to check my email I’m going to use tools that don’t allow me to open social media I’m going to set Focus time during this and I’m not going to say what I’m

(26:02) necessarily going to do but I know I got four hours from 12: pm to 4 pm on Tuesdays that I can just Vibe you know here’s my jazz time and you’d be amazed by what you can do with that because you don’t have cognitive load at all you know you just hit lunch and then you’re free and then you you can just

(26:26) you’re free to wander and if you have a project that you’re working on you know you’re undergone and it’s it’s Friday and you know it’s due Wednesday you don’t have the mental load over the weekend of saying I have to work on this or work late because you know you have a solid four hours where you’ll get more

(26:42) done than anything ever before because this is the time and you the but the trap with that is getting to a point where you’re like all right I’m going to dump everything in that time I’m just going to leave it for that because it’s not for that it’s this is your medit ative State this is Romo Santiago from

(26:58) experiment Nation every week we share interviews with and Conference sessions by our favorite conversion rate optimizers from around the world so if you like this video smash that like button and consider subscribing it helped us a bunch now back to the episode um Sean we were just talking about minial tasks and the fact that

(27:14) they create mental load for the team I want us to elevate from the work and actually think about the long-term in company level other do these menial tasks and the the mental load and fatigue that comes with it can affect our company culture over time so the real culture of a company is how effectively it listens to the pain

(27:37) points of its people right so you can’t build any type of culture that doesn’t work that way it is a culmination of a bunch of different things and a bunch of different just energies so the way that you can identify what a culture is is by how the employer looks at the employee and says okay how can we not have you feel like

(28:08) you don’t have locus of control locus of control is psychological concept that determines do you feel like you are internally or externally motivated in the things that you do so do you feel that you have control over the decisions that you’re making or do you have someone else that’s telling you this is

(28:27) everything that you have to do you know the time tracking Big Brother Big Sister looking at you type thing menial tasks by definition don’t need a human to be done so if your job to be done is high level cognitive thought and the tasks that you’re continually assigned are jobs that do not need high level cognitive

(28:55) ability that is an immediate red flag for everyone about the quote unquote culture of the company okay a toxic company I hate that because it’s overused but a company that will definitely be thought twice about is one that looks at you as a cog and a wheel and does not concern itself with your mental connection and

(29:28) well-being with what you do okay Beyond like wrote stuff like bathroom breaks and and and pay well and all that stuff right that’s to me table Stakes of a company that that deserve to be having employees in the first place okay mutual respect that’s but what I mean is does an employer look at a situation and say

(29:52) hey I’m hiring you for an output by way of renting time for you so that way you have a security level so that way you can be free to put whatever time you want to input because I need an output from you and during that time I don’t necessarily necessarily care what it is that you’re individually doing as long as the output is there for

(30:17) me because I’ve understood that I’m purchasing an output for you and and I’m not necessarily renting your time it just depends on on the the the NBA is running it like are they looking at you as a commodity or are they looking at the overall holistic view of the system and understanding that your input is the thing that

(30:36) actually matters so it just depends on how they look at it but menial task regardless of how you are assign them by definition will remove locus of control from you because you understand that they don’t have to be done by you but you’re doing them because you’re told to so if your dad is giving you an

(30:55) assignment to take out the garbage you just do it because you’re told to he easily could do it this is menial task menial labor like there’s no literally no reason for him not to do it other than you exist okay so that hierarchal nature doesn’t doesn’t have the culture of breeding a communicative like um way of moving

(31:21) forward all that is is get in line by behind me and do what I tell you to do and here are the tasks for you and this is the way you do those tasks and don’t step out of line and if you do someone else will be here to do those tasks because you don’t have to so it’s not in any way human first it’s systems first

(31:45) so it’s the opposite way that I look at things so I want to remove the medial tasks to allow the human to be human and other people want to um remove the menial task from the Human by removing the human so they want to figure out a way to identify these menial tasks which don’t have to happen at all but also um

(32:16) identify the steps that the human has to have and then automate that too so it’s the difference between um Henry Ford and Jeff Bezos if that makes sense Henry Ford understood that he didn’t necessarily need skilled labor to build a car he needed semiskilled um wheel Baker makers and he needed semiskilled um engine builders like

(32:43) these are very skilled specifically skilled people but each one of those people had a specific Ro and those people had a line in the assembly now BAS looks at the human and looks at that as training data for the AI so Bezos and Elon Musk they look at this as yeah you have to exist as inputs to the system

(33:08) because my systems need to learn my systems need to understand where you take these menial tasks okay so the culture in those businesses are optimized immediately all things everything transactional make things go faster doesn’t matter how it happens I don’t care if you have to pee in a bottle go faster systems first Elon

(33:34) Musk and and and um Jeff Bezos where Henry Ford is definitely systems first but he understood that the humans were the critical component of it and that’s why every other manufacturer and every other industry were forced to do what he did because he said I’m going to pay you a living wage to work in my factory to become a skilled

(34:00) laborer and you’re going to just do this for the rest of your life so he changed the game and that’s that’s what I mean by he didn’t go out and decide to build a culture of Automation and build the culture of treating his employees right that’s not how it works Ford looked at the problem and said I need to build

(34:19) cars I can’t get people to build cars because it’s too complicated break this down into individual systems let me get people in here that I can train and pay higher than anyone else and give them locus of control not because I want to not because I’m for was a horrible human not because I want to be because it’s

(34:37) best for the system this is the critical thing he didn’t really care about the humans at all but he realized if he got out of his own way and made it better for the humans as an input the culture would be there and it would build itself and he would become the richest person to do this and he would change the way

(34:56) business is done and it wasn’t altruis and it wasn’t because he cared about humans or because he wanted the he wanted the best product in the best way that was repeatable and this was the only way that it would work it just so happened that it allowed things like unions and stuff like that to happen um

(35:09) that’s a knock on effect you know it’s like Bill Gates exists and he’s kind of really a horrible human but that begets Melinda Gates and we get malaria being solved you know what I mean like Jeff Bezos exists and he’s kind of a horrible human but that begets the him divorcing his wife and she gives $650 million to

(35:32) you know women funding so it’s like the the culture of all of this can’t be built it just it only could be recognized after the fact you know so but how do you do it how do you build how do you place it in in the building blocks in the foundation of how to do that is to understand that as the employer you’re in your own way if

(35:55) you’re if you’re if you’re just filling their time you know if you’re just giving them because you think you’re renting their time and if they’re not busy then it’s inefficient um that you’re you’re kind of just cutting your nose off despite your face at that point you know so like there’s no real right answer to that

(36:14) there’s no way how do you build the culture you can only set it and then at the end of it be surprised and look back and say oh yeah here it is you know those examples were very evocative me um here’s my my last big question we see that right now there’s always a very hot Trend right now if we go on LinkedIn and

(36:35) open up 10 posts we’ll see AI come up a lot there might be a different flavor of the month later but what advice can you give teams and companies when they’re looking at these different Technologies and tools and getting excited for them to avoid the pitfalls before getting too deep into something that’s hard to get

(36:54) out of H well I knock no hustle okay so I don’t speak of any program or software person or anything ever so I celebrate every single human that’s out there trying to wrap around GPT and to trying to build something um so yeah the thing I would caution against though and this is the the problem I think

(37:24) is just like any anything else you see this in crer often if it’s too good to be true it’s just not true I don’t know how else to really say that and there’s no real like concrete examples because maybe and honestly I hope every single person that says that they’re that they’re using Ai and the way that they’re using it um is genuine

(37:46) and they really believe that they’re doing that and I that’s how I live my life and I take them at their face value and I don’t hate on anyone and I and I celebrate them until they give me reason then and then they become irrelevant but what I would say though is anyone that tells you that they have

(38:03) something that is nonspecific meaning it can do more than one thing what we have right now is what’s known as specific AI okay so for example let’s say we taught an AI to build or write the absolute best Dungeons and Dragons or Lord of the Ring fanfiction I mean this thing is tolken JR okay it it wrote that it wrote you

(38:32) know the the next version of this and you couldn’t tell or Game of Thrones pick an author pick any pick an author you know let’s say Jr Martin Dies or whatever like Georgia Martin Dies and he never finishes his book we get an AI to finish it and you never know that he didn’t write it okay in that instance

(38:53) that’s first of all right now I wouldn’t say impossible but it’s going to take it’s going to take some time and effort and it’s going to take some some really smart some really smart stuff but it’s possible you could do it okay cool we’re in that world right we got we got Jr token or George Martin AI it can bang

(39:17) out um manuscripts or or books in their in their voice that is indecipherable from anything that they would have written themselves great awesome tell that thing to make you a recipe for baking a cake tell it to write you Pros tell it to create you a sonnic tell it to do any other lit literary tasks that would be within the

(39:43) realm another example let’s say we’ve created it’s an AI Mitch Hagberg and we’ll use a mitch Hagberg joke from this he famously said comedi Ians are the only profession in which being successful in it people will ask if you could do another profession oh you’re a comedian can you act okay if you built a

(40:04) comedian AI right now and it was the Best comedian first of all it would have to be a singular comedian it would not have to be it could not be a comedian it would have to be an AI Mitch Hedberg an AI um George Carlin and AI pick one doesn’t matter you can then train that and then over time make it indecipherable from this person great

(40:29) have that comedian then go right a play it can’t so that example and the reasons I go into that is so you understand that we have at the most specific Ai and it’s not even AI it’s llms so we have specific L llms that give you the statistical answer to a question within a frame of reference this is not anything that is purported

(40:59) by most of these tools so most of these tools that will say hey we have we have ai or I am an AI tool they mo around it is basically none because what they are is a wrapper around an API of GPT and it’s it is a very clever smart and efficient prompt Builder so the problem with that is it’s like building a WordPress

(41:36) plug-in when WordPress just takes the best stuff out of every plugin every built every three months and incorporates it into WordPress and you’re building something that is reliant on WordPress still being 63% of the internet or whatever number it is right so you build this amazing thing and you’re like oh man you got all

(42:00) these users and you’re like this is great now we have this thing for WordPress and it does this awesome amazing thing and now WordPress says oh that is really dope hey WordPress version 7.5 now includes this and because it’s open source and because the way you built it and because you built it on the backbone of a thing you’re you

(42:18) don’t have a moat there like it it’s just anyone could just grab it and if you’re building something just to get your name out there and just to get you know your your uh your chops on AI That’s great but if you’re on the other end of that and you’re and you have one choice to purchase a tool and the person

(42:38) that is building the tool is doing it because they believe in it I do I agree with that but also because they they need to do this they need to get their reps in and they’re doing a rapper that they that they inherently understand will probably just be relevant in 3 to six months another example would be

(42:58) anyone that spent the last year or so building an AI Video Creator off of images and text that can then Stitch your your your V your um images into video those people who have done amazing work and have really put things forward and us it like stable diffusion models and really gotten to a point where you

(43:16) can create like almost pretty cool video you know like then Sora drops and now what and now what you know what I mean so yeah if you were one of the people that that did an RFP or you did one of those people that that paid 20 grand to get this thing on your system in your Enterprise and you’re looking at it and

(43:39) you’re like well this Tool’s now garbage or you have maybe you have an AI personalization tool that you’ve been running and you realize that it’s not learning the way that it said it was learning and now they’re pivoting into using an llm and you’re like but wait I thought we were training all this data

(43:54) and they’re like yeah just kiding we realized it was going to cost a billion dollars to be able to build the context engine to run this actual data set that we’ve been training on and it’s no fault of theirs because they don’t really think about that they don’t really think about yeah great let’s train let’s let’s record all of the data

(44:14) that you’ve done let’s record all the personalizations that we’ve done let’s record of maybe the experimentation outputs that we’ve done that’s awesome but there’s a very big step that has to go between that training data and an llm or an what we call an AI and it took like I don’t know 150 $200 million

(44:35) from 2015 to 2021 for open AI to even announce gpt2 so yeah sure if you have that kind of time and that kind of money to have somebody sit there and human human go through this and do all of the stuff that you need this is the thing you don’t no one does and anyone that tells you that they’re going to

(45:04) train on your data or they have data from previous that they’ve trained on no they don’t because they didn’t know what data they would need to train on the system that they hadn’t built yet so that in itself is a paradoxal conversation where they say hey we have an AI that we’ve built in the last year

(45:24) that’s being trained on previous data from the last say six years because we have this huge data set but there’s mental gymnastics that have to happen there and we just have to eliminate the fact that this AI that there was no way for you to ever understand what you were going to be recording at the time you

(45:43) were recording it in that data set for this so you have to shape that data somehow you can’t just plug an llm into it and be like hey figure this out it doesn’t work that way so you have to take the time to have a human go through and do old school natural language processing sure sure there’s machine

(46:05) learning that happens here but this is different this still requires a lot of work and I don’t know that the people that are saying hey we have this Ai and they just wrap around GPT have done this work so lesson is be skeptical look at tools that are mostly focused and anything that looks too good to be true

(46:26) be wary of it and dig deeper investigate I mean that’s tool for that’s true for any tool right sure is I mean that’s tool for any marketing tool for any LinkedIn tool for anything anything that somebody sends you an email and says I’ll make your life my own tools like my automate to win stuff you should be skeptical about

(46:45) because I’m just some guy on the internet telling you that I can remove the busy work in your life that you should be skeptical and I should give you enough proof that it comes to you you’re like yeah that makes sense because if you just accept it then you’re accepting the fact that there’s things in your day that you’re doing

(47:02) that you don’t necessarily have to do and that’s a trap too because maybe there’s not maybe you’re maybe what you’re doing needs to be done and you just don’t like it and that’s fine and I understand it but that’s different than saying this is busy work you know like this doesn’t have this a robot could do this only because

(47:25) you don’t enjoy it because you don’t or not necessarily you whoever doesn’t doesn’t enjoy the process that doesn’t necessarily mean the system is bad or the process is bad that just means that you don’t enjoy it and that that’s a possibility too the possibility is you’re just a bad fit in the system and

(47:41) you need to move in a different position in the system and that’s why human first is so important because it just doesn’t matter about what you’re actually doing like especially if you’re renting time your time is already rented so just you can figure out a place to fit where you can get Locust of control back you can

(48:03) actually start to feel like you’re contributing and then the things that you were feeling were busy work other people could do they don’t feel like they’re busy work because they feel that it allows them to release whatever they need you know so maybe one person doesn’t like to be in task management

(48:18) and doing that type of scheduling and they feel like that’s busy work and it’s like beneath them or like it’s just constrictive for them and another person is super organized and super task and loves lists and gets a dopamine hit when they get to move things to status to status and their whole entire life is

(48:35) super organized just switch those people and put the person in charge of you know organizing and running the system where they enjoy it and get the other person out and you automate out both of those people’s systems so the person that doesn’t like like reporting out and communicating to other people because

(48:55) their systems focused and maybe introverted what you do is you communicate out so you automate the systems there that says when I do all these neat little systems things that I do build communication layers that send it out to the people that don’t care about the systems things and the other side is when I do cool Communications

(49:12) things like when I build a new asset or when I push a new git commit or I you know close a new contract or I or I create a new contract in my CRM that is updated within the task layer so the person who loves the the levers and the buttons gets updated so then they can move the levers and the buttons which

(49:35) then gets sent back to the person communicated so you never like remove the systems you just remove the part that they don’t want to do and that’s why it’s human first because every person’s automation is going to be slightly different and there’s no like set automation so if you know Simon and Sean work at the company

(49:55) my automation set and your automation set will be different because I’m going to remove certain problems for stuff that you don’t want to do and I’m going to remove stuff that I don’t want to do but at the end of the day the building of the culture means the employer allows the employees to not do the things that

(50:12) they don’t want to do because they understand they’re getting an output and they’re not just renting time that’s awesome the the kind of outcome of all of this is eliminating all the the work that people don’t want to do granting them more quality time allowing them to get into this Flow State and the theory behind all that

(50:33) right is that we’re GNA create a much more productive environment yeah that was hope incredible conversation Sean and really grateful we had you as a as a guest here on the experiment Nation Podcast where can people find you if they like this conversation and want to hear and see more of you uh LinkedIn uh

(50:51) definitely u i just launched this this automation quiz is automat automation quiz it’s free it’s um it’s a 75 page PDF that will come back with a customize uh identification of what you can automate how to do it steps test identify like it’s a really cool thing that I put a lot of time into um yeah

(51:13) I’m I’m you know I’m on the experiment Nation slack so you can hit me there and yeah I’m I’m you know I’m online so I am here awesome thank you for joining us Sean this was an awesome conversation and I hope people took away a lot out of it but we can definitely see how the human the processes and the systems kind

(51:32) of all connect to each other and our goal is to deliver you know higher quality work and higher appreciation of our time spent working so this was an incredible discussion and thank you for joining us now my pleasure I’m happy to be here this is Romo Santiago from experiment Nation every week we share

(51:49) interviews with and Conference sessions by our favorite conversion rate optimizers from around the world so if you like this video smash that like button and consider subscribing it helps us a bunch

If you liked this post, sign up for Experiment Nation’s newsletter to receive more great interviews like this, memes, editorials, and conference sessions in your inbox:

Connect with Experimenters from around the world

We’ll highlight our latest members throughout our site, shout them out on LinkedIn, and for those who are interested, include them in an upcoming profile feature on our site.

Rommil Santiago