115: AI for ADHD Entrepreneurs with Stephanie Ward

Welcome Back, Stephanie Ward!

[00:00:00] 

Laurence Pratt: Hello and welcome to the ADHD Goals podcast. Now I think we have got a first today we’ve got our first returning guest, and it is Stephanie Ward. She is of the, spicy Brain Collective, and she was on, a couple of months ago telling us about, how she helps ADHD entrepreneurs and, neurodiverse. Entrepreneurs with their online businesses So go and check that episode out if you haven’t listened to it already. and she is back because we want to talk about ai, but also we’ve just been chatting before we hit record and a lot has happened for Steph since we last spoke.

So, how are you today, Steph?

Stephanie Ward: Better than I’ve been for a while, I think. ’cause they’re probably the best way of describing it. Yeah, everything’s [00:01:00] been a bit chaotic. Might have run away with the circus, but I am back now.

Laurence Pratt: Yeah, I mean there was quite a few, exciting is maybe a word I would use. things that have gone on. I just wonder how many, how much of that you wanna share with the listeners to just keep them up to date.

Stephanie’s Eventful Summer

Stephanie Ward: Yeah, I think I’ve had the most ADHD summer ever.so I did this really cool business project up in Scotland with a very old friend of mine. very fun. I came back from that with some quite profound realizations. I think when we get to sit in silence and let our brains process, I actually got a nap during the day ’cause we were doing like bushcraft and I don’t do outdoors, so I had a daytime nap.
So I’m very much relaxed in a deep level and I realized that things at home weren’t quite what. What I would’ve expected at this point in my life, and there was a moment where I realized that I’d actually learned to be helpless in my personal relationship with my partner. And I decided on a whim, just one day.
There was a one minor. Conversation. And the next thing I know, I’ve broken up with the man I’ve [00:02:00] been with for 12 years

Laurence Pratt: three hours later I booked a flight to San Diego to spend a week with someone I’ve met twice in real life. a guy called David Fagan, who’s a talent agent for top talent. And I just spent like a week just going round events.

Stephanie Ward: I went to an event with the fabulous Lisa Nichols. If you haven’t seen the things she does, go and check her out. She’s wonderful. I ended up going over to Temecula, say a friend Diane, who was filming a TV show that day. I hung out with my friend Helmut, who was gonna be the mayor of Atlanta in Georgia, just ’cause he randomly happened to be at the same event.
and I’ve got really good friends, a lady called Melanie Soay who does humanitarian aid trips in South Africa, and it turns out she used to be the deputy District Attorney for Los Angeles County. Right. And she’s just this Feral woman who just she’s like the pied Piper. If Melanie says, come, you follow, and you don’t ask too many questions about it.

So we just had the best time. We went to the Grand Canyon National Park and sat on a cliff 400 feet in the air. yeah, it was just the most relaxing thing I could have done. so I’m back from [00:03:00] that. I’ve been at a few conferences and events since, because that’s. How I move. But yeah, breaking up with my boyfriend and running away with a circus was not on my 2025 Bingo card.
So I’ve got a lot to figure out right now ’cause I’ve realized I’ve got no game with boys. So that’s a whole situation.

Laurence Pratt: being in the news at all,

Stephanie Ward: Oh yes. And that. 

Breaking News and Opinions

Stephanie Ward: so yes, after I got back from the us, I had seen some rumblings online about our lovely friend Donald Trump and his stance on paracetamol causing autism. and I got a message from Metro News on the Monday saying, do you have an opinion on this? I was like, do I have an opinion? Yes.
so I wrote a piece the day before it broke. Essentially with my views, but I had to stay up for the live broadcast because I wanted to make sure what I said reflected it. And honestly, I was horrified at what he was saying because not only has he said that something is harmless as paracetamol, which has been repeatedly shown in both independent studies, sibling studies, peer reviewed studies, not to cause any kind of neurodevelopmental issues whatsoever.

He also revived all of the [00:04:00] Andrew Wakefield research around the MMR vaccine. That was around sort of the 19 98, 19 99, timeframe. And it’s the reason that my own brother wasn’t vaccinated, because my mother thought it would cause autism.

Laurence Pratt: and I think it’s a very damaging narrative to have out there.

Stephanie Ward: And I, my personal stance is stop trying to find out what caused me and just celebrate the magic that is me. Because I think the autistic brain and the way the condition presents in people can bring some really powerful. Useful skills to the workforce as much as anything else, but it makes some really interesting people and I just really wish that they would stop trying to find a cause for it and just let us live our lives, stop treating us like some health crisis and just let us be the beautiful humans we are.
So that was in Metro? It was online and it was also published twice in print. So I’ve got it on my wall now. I’m so excited. I had most of my friends trying to find the paper that day just so I could have a coffee.

Laurence Pratt: Yeah, I think you know that paracetamol thing as well. I mean there’s also another way to.look at what they’re doing there as well. and I think it, it goes along [00:05:00] parallel to sort of that traditional family thing that they’ve got going on where, a woman should be, there to give birth and stay at home. and, taking things to make, pregnancy easier is bad ’cause you’re gonna give your,child autism. I think that’s another thing that they’re sort of. Sort of pushing as well, and I think that’s, yeah, it’s all a bit scary, I

Stephanie Ward: Yeah. It’s, they’re just, basically what they’ve done is they’ve created another stick for society to beat women with is what they’ve done.

Laurence Pratt: already torture ourselves enough during pregnancy because oh my goodness, is being in the car too long gonna mean I inhale gases and damage my child, or I can’t eat blue cheese ’cause my child might die and I can’t do this because I might sabotage my pregnancy and I can’t exercise because I might tense too hard and break my waters like.

Stephanie Ward: We are scared enough during pregnancy without them taking away. The one thing we can use pain relief because pregnancy is not very comfortable. I dunno if you’ve ever carried a bowling ball around on your crotch, but that’s what it feels like. I [00:06:00] had a large child and my hips went at. 26 weeks. So I was on crutches on, actually, I was on tramadol when I was pregnant because I was in excruciating pain for nearly three months.

I had to have an emergency C-section because I, it got to a point where I was in so much pain I could barely get to the bathroom and.I started the Tramadol around for the 32 week mark. Before that, I was only on paracetamol and that was the only reprieve I got was my four times a day paracetamol. And it didn’t make my son autistic.

Do you know what made my son autistic? Me and my ex-husband who can tell you what A BMW is, even if it’s de badged based on the allo wheels. So I feel like that might be the reason they’re not the paracetamol. Yeah.

Laurence Pratt: it’s not as if, it’s also suggesting that being autistic is wrong or something, that if you had the choice not to be, you wouldn’t, it’s no, it’s a different way of being and that’s fine.

Stephanie Ward: Yeah.

Laurence Pratt: it’s just more a case of let’s make sure the society we live in room for everybody.

Stephanie Ward: Yeah. And there’s more than enough space for all of us to be [00:07:00] exactly who we are. Stop trying to find the cause and spend the money on putting the right adjustments in so that we can be the valuable members of society that we already are, that we’re just not celebrated for.

Laurence Pratt: And so as a segue into the topic of today’s episode, there are some tools that we could use that, in this society that we live in, that the expectations of what people should be able to do at work. we’re in the neurodiverse space there’s lots of things that people can struggle with, and has the,the possibility to help. Let’s do that. and that’s what we wanna really, sort of look at today. 

AI Tools for Neurodivergent Entrepreneurs

Laurence Pratt: So I wonder if, you could sort of give us a bit of a broadof what you are finding interesting about AI at the moment and, what you’d like to talk about today.

Stephanie Ward: Yeah, I mean, so in the last year we’ve seen a lot more people who are getting things like access to work. Work that are getting funded for stuff [00:08:00] like, AI based note takers, for example, or they get given things like, Clickup, which has the AI built in most platforms that we are using now has built in AI or AI capabilities.

And what’s really cool about this is that. When it comes to neurodivergence, we struggle with certain tasks, and my whole theory around this is it’s not the task that we struggle with on its own. It’s the first and last 10% of the task. It’s the getting started, getting the structure, getting the confidence, but then also the closing the loops at the end and.

What I find really good about AI tools, things like chat, GPT, although I have mixed feelings on open source language models as the main source of AI that we use, but things like, you know, Sintra or copilot, all of these different things. What I love about them is that you can give it information and say, give me the, give me a starting point.

And for neurodivergent person who’s stuck in task paralysis or struggling with executive dysfunction, what it does is it gives us that starting point. That means we can do a lot more. Autonomously it’s things like, you know, giving us the [00:09:00] bare bones of a social media post. So if we’re struggling to out what will go well on LinkedIn a day, we can ask it and it can say, here are some things you could post about.

Would you like me to draft a post for you? Yes, I would draft me one, but what worries me is that. I think there’s a lot, there’s a lot you can do with ai, right? You know, we’ve got AI agents, which are gonna be the future if I’m honest with you. I think a lot of monotonous tasks do not need to be a human.

There’s a lot of things that an AI can look at and be like, do that. And if you give them the boundaries and the parameters, you can train that like a, like an employee. You can give part of your brain to that, and it just takes care of that one task really well. But the problem is that people are becoming too reliant on ai.
They’re using it to formulate the entirety of something without actually, validating whether it’s true.

Actually fact checking what the AI’s putting out, what I’m seeing right now is that autistic people are using it to get facts, to get information. We know that the youth now aren’t using Google to find things.
They’re using generative engines [00:10:00] such as chat, GPT. Which is why a lot of companies are shifting their focus from search engine optimization towards generative engine optimization. So when you go to chat GBT and say, give me the five best business consultants, it’s people who’ve done their GEO that are gonna get burned out of that.

So we’re not actually fact checking whether they are the best. What we’re doing is it’s pulling information that other people have said, this is the best person, this is all the good things about them. So it’s just as biased as getting an opinion from another human in some ways.

Laurence Pratt: So I think, what you’ve touched upon there is something I probably want to just check in with the audience and make sure what we. 

Understanding AI and Its Applications

Laurence Pratt: What we mean and understand by AI, because it’s a booming topic at the moment and everything has got AI in it. The word AI in it, because everybody’s gonna think you’re gonna click on it if it has.

and what you’ve sort of highlighted there is, all the tools that we use already ha are. Integrating AI into it. We’ve probably all used chat, GPT. So when we say ai, we’re probably thinking about that. But then you’ve [00:11:00] also mentioned that there’s agents and those are things when I think about agents, I’m thinking about those platforms like n8n, where you can.

plug in agents and get it to do plug in different tools and remember stuff and it can do things autonomously. and that’s, there’s a really broad scope. it’s like saying, what do you think about religion? And there’s so

Stephanie Ward: Yeah.

Laurence Pratt: the what that one word in the person you say it to that instantly triggers them to a point of view of that they have, whereas, and you can’t have a conversation about it. If the person’s got another point of view. ’cause it’s too big a word to talk about in context of what you mean. So I think it’s important to sort of scale it down to what we mean by AI and maybe what that, the journey of tools and possibilities and leveling up what, what we can mean by ai.

Stephanie Ward: And I, I think as well, a lot of people don’t have a full understanding of what AI is and the different levels of it. I think people say that they’re expert in ai. Well, [00:12:00] actually what they’re an expert in is crafting a prompt for chat GPT,

Laurence Pratt: Yeah.

Stephanie Ward: and from an entry level perspective, chat, GBT is the most accessible.
That’s the bottom level. That is something, it’s a chat bot based thing where you can go and ask it a question. And the open source language model gives you answers based on its knowledge base.

Laurence Pratt: And that’s really good in terms of, it’s a bit like Google, you know, we go to it, we ask it a question, it gives us an answer.

Stephanie Ward: it can steer us a little bit. It can, you know, based on what it learns. What’s really good is that open source language models learn and learn and bring more information. So you might ask it a question now, you might get a different answer in six months because it’s learning all the time.
Whereas things like the agent based stuff, so what Jeanie McGilvery is building over Autumn AI is agent based. It’s very specific things, doing very specific tasks on very specific platforms with not a lot of scope for, learning from external like models. It’s, you’ve given the information, you train it on what you know, it’s given parameters around specific tasks and it does something specific.

Laurence Pratt: Yeah.

Stephanie Ward: And [00:13:00] I think when we think about AI uses for Neurodivergence, they don’t tend to be using things like the agents. What it is things like chat, GPT or RA, or Google, Gemini. It’s the open source language models where you can have a conversation with it. And I think the problem is that we all.

Neurodivergence, we know that part of the neurodivergent profile is struggling with communication,

Laurence Pratt: Yeah.

Stephanie Ward: a lot of us prefer a text rather than call because you can have a chat over a text message much easier. You can formulate your thoughts and I think there is this element of, it feels like we’re talking to someone, so it feels like we’re getting more social juice as well.

So I think that’s why people go down that, that route. And why that’s probably the most common uses of a usage of AI in Neurodivergence. Is that because we like the chat aspect.

Laurence Pratt: Yeah, I think what you’ve sort of. Highlighted there. I, it, talking about the agents as well. I think, just to throw another term in there, I would say automations, when we talk about automations have been around for ages. and for anybody who’s not [00:14:00] sure, I mean by that, like think about, you do a, an email campaign, a MailChimp or something, somebody downloads a lead magnet and then you send, send this email. After one day, two days, send this email if they reply, if this, that, and that, that is an automation. And that’s been around for ages. and essentially what we’re sort of driving at with agents is saying, how can we specify an automation, I want it to do this through, to this, this workflow. and. If you imagine like you, you might have a chat bot for example, and when you do that with an automation, you probably give it like two or three. Answers to give. That was how you would do it up until before ai and you’d have to think about what the possible answers are and that for someone with ADHD seems like a big, massive headache.

And probably, there’s probably a lot of people listening that have got automation software, but just really can’t be bothered [00:15:00] To map it all out.

Stephanie Ward: It’s a huge job to map these things out as well though, because apart from anything there’s neurodivergence. we don’t see things from anything other than our own frame of reference. And it’s not the fault in our program and that’s just the way we do it. So how can we possibly anticipate people’s questions to the extent we need to when we don’t, we can’t see the questions ’cause we know the answers, so why would anybody ask that question?

So it does require like a lot of sort of help to actually program it in the first place.

Laurence Pratt: Yeah.and yeah. And I think that is the key. and I think that anybody thinking, people probably clicking on ads and seeing things on, social posts that, you know, LinkedIn, that AI can. run your business. It’s only gonna be able to run your business if it knows how you run your business. And I suppose a lot of the big, the common trap that a lot of people with ADHD who are running their business is that only exists in your head. And if you can’t tell somebody else how you run your [00:16:00] business, you can’t tell AI how to run your business. So I suppose, it comes back to that, that process of. Creating SOPs in your business. If you can create an SOP, you can then train an AI on that, and have a lot more success than what we, a lot of people might be doing, which is just day they go on, start a new chat, ask it something, and then chat. GPT. just wants to impress you by keep going, shall I do this?
Shall I do this? And it gets so far away from what you originally wanted to do that you might be forgiven for thinking, well, I’ve tried check GBT and it’s rubbish. And like you say, you’ve got reservations about it, but it can be a lot better if you are a bit more strategical about how you use it.

Stephanie Ward: Yeah, absolutely. 

Creating Custom AI Solutions

Stephanie Ward: So one of my clients are doing something really powerful in terms of building chat bots. So she’s autistic with ADHD, but a lot mort is than the ADHD. So it’s interest to see how that balances out in different people’s profiles because she’s able to capture [00:17:00] stuff. But the a d ADHD is a memory issue.

So what she’s doing is with her support worker is building out these SOPs guidelines, rules around the brand to populate chat bots. specific things. So she’s created one that is a problem solver. So when she goes to and says, hi, this is the problem that my organization is facing today, based on what you know about us, how would you go about this?

And it’s got all kinds of things like, about how they like to do their customer care. How would they like to, post on social media what their brand voice is? So if it’s gonna draft an email, it sounds like her the first time.

Laurence Pratt: it’s things like if they’re doing fundraising bids, there’s a separate one for fundraising bids that knows all about their objectives, all about their past successes.

Stephanie Ward: But then it can say, right, okay, you’re writing for this one. Mention this here. And it gives the right answers. But it does require a library of information to work from because if you don’t give it a library, it’s just gonna pull from the open source language model,

Laurence Pratt: you could get absolutely anything out of that.

Stephanie Ward: And that’s quite scary.

Laurence Pratt: Yeah. And so, so just thinking about. We mentioned agents, but I think we can get, we’ll see if we’ve got time [00:18:00] to get onto those. we can towards the end, but I think we can get a lot of practical tips. this conversation, just talking about what can be done in something like that. We’ve probably all got, which is chat, GBT.

So about those you might have or some documents for knowledge base, I wonder if we could talk a little bit about, Going beyond that, just click and chat and start maybe talking about what a custom GPT is or even a project folder and how we can use those to, more specific and more useful.
We chat GPT.

Stephanie Ward: Yeah, so I think the first thing for a neurodivergent person is it’s great to have your own custom sort of GPTs having like folders via your product, all these different things. But there are people out there who’ve already done the legwork to get you started. You do not need to start from scratch with this.

There’s people like Darius Lucas, for example, who’s put together a whole list. Of starting points, like social media marketer, all these different things. So they’ve got a lot of the background knowledge. You just need to feed it about your business. It’s got the background knowledge to [00:19:00] know about the platforms and to know about all these different things.

it’s an absolute waste of anyone’s time to try and build that from scratch. Scratch, because you can get your hands on or people’s starting points for free or very cheap.

Laurence Pratt: Yeah.

Stephanie Ward: Don’t waste your time trying to do the first 10% of this.

So in terms of, so let’s just sort of, so a, for anybody that doesn’t know, like a custom G-P-T-I-I is basically a chat, like it’s a little. Chat that you can create that has sort of guardrails in it. Yeah.

Laurence Pratt: you’ve got the premium account, you can create your own chat GPTs and you’ve probably got access to everybody, marketplaces like the ones that you’ve just said.

and that process of creating a custom GPT is uploading documents it that tell ’em about your business. But then, and then you know, specific things about what the process of what it wants to do. Like an SOP,

Stephanie Ward: You have to treat it like a 5-year-old. Essentially, what if you are gonna do a custom GPT? You have to treat it like it’s a five-year-old that doesn’t know [00:20:00] anything about the thing, and you have to give it all of the information that it needs to do a job like you would an employee. but further back when you get an employee, you assume base knowledge, but with a custom GPT, it’s pulling from so much and also nothing at the same time.

So you have to be very specific about your business and it’s having these knowledge files. If you haven’t got SOPs, it’s gonna be really hard to start that because you need to have that base knowledge captured somewhere to feed to it. And it’s a big job. It’s a worthwhile job if you do it.

But it is a lot of work to do from scratch and, there are benefits to building it is the first thing you know if you are gonna do fundraising, it is a great idea to have a custom GPT built around what your objectives are for that.

Laurence Pratt: Yeah, so I was just gonna say if, I mean, I say SOPs a lot. I think we probably said SOPs a lot in our last episode, but for anybody who’s, doesn’t know what that is, it’s like a checklist of how to do a specific task.

Stephanie Ward: again, for someone with ADHD creating one of those sounds like a massive headache.

Laurence Pratt: But, I, [00:21:00] you’re probably familiar with the tool scribe.

Stephanie Ward: scribe right where the click by click.

Laurence Pratt: So

Stephanie Ward: I love that.

Laurence Pratt: you just sit there and perform the task, and it’s taking screenshots as you do it. And then at the end of it, it sort of annotates it up like someone has just made a really well thought out, detailed, SOP. So there’s a shortcut to creating those horrible really important documents.

Stephanie Ward: There is, I think with things like that, like Scribe, the day I discovered Scribe was one of the best days of my life because a lot of my clients have memory retention issues. So if I show them a task on a call and said, scribe away, it was documented straight away. You could use that as a starting point and feed it to chat GPT, for example, and say, can you turn this into a written SOP or a, you know, if it’s, if you need to have a policy around it, for example, and you can write your policy up based on what you’ve done.

So if you are struggling to get these SOPs in the first place, use something like Scribe to get your guide and then get it to write the policies around that. Like you [00:22:00] can use it like have you come across Notebook, lm.

Laurence Pratt: I don’t know. Oh, excuse me.

Stephanie Ward: Okay.

Laurence Pratt: is that?

Stephanie Ward: So basically Notebook LM is something that Google brought out quite recently, which essentially you can create a knowledge base

Laurence Pratt: by feeding everything you have into Notebook, lm. But what it can do is it can actually have two voice models talk about the information you’ve given it. So it’s like almost like it makes a little podcast out of your stuff.

Stephanie Ward: So if you are, if you say a bit like, tell me everything you know about this thing that our company does, it sounds like people are having a conversation about it. But what it’s really good for doing is highlighting gaps.

Laurence Pratt: Yes.

Stephanie Ward: So if you don’t know your blind spots, you can feed it stuff and it’ll start talking about what you fed it and they’d be, oh, it’s missed that, and it’ll help you to fill in gaps.

Like I genuinely have a lot of love for ai. I just feel like so many of us aren’tgetting what we need out of it.

Laurence Pratt: that, that’s it. I mean, it’s good for, you know, that blank page and if it does. Produce something, it can sort of, it can sometimes be better at, make making [00:23:00] things more concise than you would say it. But if it gets something wrong, then you can say, well, actually no, it’s not that.
And you can correct it as long as you take the time to sort of review what it says and check it and don’t just send or use whatever comes out. I just, I mean, I do, talking earlier about, it’s possible to, make shortcuts and there’s lots of tools. out there. would, a lot of the fun that I’ve been having is like looking at how to make things from scratch. and one of the good things that I found to, is, to, give the AI a role and literally to say. You can say, look, hey, chat. You are an expert AI engineer, you are, I am your client and I want to build this. What questions would you ask me? If you ask it? What questions should it ask me? it will ask you questions. If you tell it to do [00:24:00] something, it will do it, and if your knowledge is narrow and not broad enough.you’re not utilizing what it is that you don’t know.

Stephanie Ward: Yeah.

Laurence Pratt: it’s always good to ask it what, you know, put it in the, in an expert position and ask it what questions to ask, and then sort of narrow it down.

And with that, like I’ve built like a, I’ve built a custom GPT So

Stephanie Ward: Nice.

Laurence Pratt: if you create your SOP, you can say, right, I want to, upload this.I’m gonna upload this and you’re gonna turn it into the documents that I need to create the custom GPT. So it comes out with the knowledge document that you put in, and also the instructions to say,this step, and then stop. And so that’s a cool way to make g TBTs, but like you say, there are a lot out there already.

Stephanie Ward: Yeah.

Laurence Pratt: and also I’ve made a custom GPT that produces.you’ve heard of the business model [00:25:00] canvas.

Stephanie Ward: Yes.

Laurence Pratt: So like the business model canvas is like a really,a succinct way of describing what your business does and what, how it all operates. And so.you can basically go and sort of look at your website and say, is this your business model canvas? And you go, yeah, that’s pretty much it. And the value proposition and everything. And then you’ve got pretty much like a document then that you can then inside your, custom GPT alongside the SOP of the job you want it to do. And it’s, that’s A whole lot more trained than

Stephanie Ward: up a chat and saying, Hey, can you produce a blog post? Or whatever.
Yeah. 

Understanding Chat History in AI Tools

Stephanie Ward: And this is the thing I think a lot of people don’t realize that when you start a new chat, anything you’ve told it before isn’t in that chat. So if you’ve spent a lot of time having one chat and feeding a lot of stuff about your things and then you start a new one, you lose all of that.

The Importance of Strategic Use of AI

Stephanie Ward: So it’s very important that if we’re going to use it, we need to be strategic and understand what it is that we are doing with these [00:26:00] chats.before we even go there, and I think this is where a lot of people have fallen down, because I think people think that, oh, it’s in my account. I’m just having a chat.

I say, yeah, but you are using a different chat to last time you’ve spent time uploading documents, all that stuff. And actually, that’s not captured in the new chat. So you’d have to either re-load it all, or you need to use the previous chat. A

Challenges with Chat Memory

Stephanie Ward: nd there was somebody on my Facebook a little while ago, very intelligent woman, very.
Successful woman was getting really frustrated with chat, EPT. ’cause it’s forgotten everything I’ve ever told you. It’s well, yes, one, ’cause you’ve opened a new chat babes. That’s why. And she’s oh. And it was just this moment of it was just totally just ADHD brain glitch where she’s oh my god.

As if I’ve done that. But it’s like from a neurodiversity perspective, I think we don’t, we didn’t do enough learning in the beginning with REI. We just kinda, oh, this is a fun thing. Let’s have a go of it. 

Balancing AI Assistance and Human Intuition

Stephanie Ward: I mean, the problem is that I think in some ways we’re too reliant on it because we’re trusted to validate our ideas.

Laurence Pratt: Yeah.

Stephanie Ward: And actually what I love about what you said there is, getting it to ask us questions because it ask, I think getting it to ask us questions that we [00:27:00] can feed it information, but then also getting it to fill in the blanks isn’t a bad thing. I also think it needs to be asking us, and how do you feel about this?

Because the chat says, oh my God, it’s the best thing I’ve ever seen. Right? If you’ve asked it to, reply to you like your drunk bestie, it’s gonna tell you the best thing it’s ever seen. So it might be pulling on old information that you’ve gone from trends and stuff. And I just think more of us need to be leaning on our own intuition, how we feel about the things we’re creating and not letting the language model do too much of that.

It’s great having it create documents for us. It’s great having it put factual information together, but we still need, if we’re using chatbot based, to actually validate that and make sure that we’re okay with what it’s producing because it’s our work at the end of the day is things that we’re putting into the world.
We need to be happy with the outputs.

Laurence Pratt: I 

Critical Feedback and AI Limitations

Laurence Pratt: sometimes as well as sort of asking it what questions it would ask me. I also, if I’ve put a lot of information into it, I ask it to be, Not critical, but I, you know, I ask it to, [00:28:00] constructive criticisms. Yeah. Can you suggest anything that I’ve missed? Or that anything that you don’t feel is, valid or whatever.

and it, and yet it still starts out with a positive. It is, it does know how to do a shit sandwich.

Stephanie Ward: I love Sandwich. Sandwich. My favorite way of giving feedback, by the way,
it.

Laurence Pratt: but it does say, ah, but you do, if I was to be critical then X, Y, and z and you are like, oh, okay. yeah,I think just un like you say, understanding that it, unless it has got, if you just give it free reign, it’s just gonna try and impress you and go off the rails. and that’s not necessarily what you want.
I mean, I don’t know. 

Project Folders and Custom GPTs

Laurence Pratt: I don’t know what the last, the latest thing is on projects. ’cause project folders in chat GPT are slightly different to a custom GPT, but there is one sort of frustrating floor with a project folder. So you can create a project folder and then a custom GPT you can upload documents and give it some instructions, but. Then you can have many chats within it. Now, it would stand to reason that [00:29:00] those different chats would be aware of each other. And at the moment, they’re not. So

Stephanie Ward: quite do what potentially it could, but I
no.

Laurence Pratt: if they’ve got any intention to make that a possibility in the future.

Stephanie Ward: Yeah, I don’t know. And I really hope they do because it’s like the project folder holds the knowledge base.

Laurence Pratt: That’s where all the information is, and all the chats are pulling from that knowledge base. But.it would make sense that the chatter would speak to each other. This is something that Sintra doesn’t do very well.

Stephanie Ward: Sintra gives you 12 AI employees, and I mean, don’t get me wrong, it’s fabulous. Like I love my Sintra more than chat GPT, but the chat bots don’t talk to each other. You populate the brain AI with the knowledge base and it. What’s really cool about RA is I ask you questions every day about your business so you can give it more information,

Laurence Pratt: the chats don’t communicate.

Stephanie Ward: So if you are having a chat with, visit the virtual assistant and then you go to see me, the SEO person, they’re not communicating, and that is a really big flaw with these, you know, multiple chats in one thing because of the fact that we’re having these conversations. Different personalities, if you like, different employees.[00:30:00] 

It’s like there’s no internet connecting the employees together if you think of it like a workplace. And I really hope that some of these models do find a way around that because that is the biggest stumbling point for me is that I need them to talk to each other and they just don’t.

Laurence Pratt: So this, I mean, I’ve thought about this as well,and this is one of the reasons why I sort of chose to sort of build a few things from by, by myself. Sorry, I’ve just hit the mic by myself, is because if you use Centerra, I’ve got Centerra as well. If you use somebody else’s. Custom GPT, you don’t really have access to the documents that go into it, so you can’t update it with stuff that has changed. And so if you have created your own custom GPT you’ve got it, you’ve got different GPT for different things. But you know, let’s say you’ve created your business model Canvas document that you put in each of them so it knows what your business is. If you are governing your. Knowledge [00:31:00] based documents and keeping it up to date, if anything changes strategically, you can update that central, that document and then upload that to all those GPTs.
And so it,

Stephanie Ward: it gets around the fact that, it talks to them in, they’re not talking to each other, but they’re updated.

AI Integration in Business Operations

Stephanie Ward: I do also think about, because. I’m going a bit too far now, but like how when you are in a company and there’s lots of people using stuff, if you have a custom GPT, everyone can use that custom GPT.but. If you have a custom GPT on your own computer that only you use, you can get it to do agent stuff. Like you can go and get it to call something if you’re getting it to do shared stuff on, like on, Google workspace or whatever. But it can actually start to do, stuff on your machine if it’s got the API to connect it.

Laurence Pratt: But that’s something you can’t really do if you’re using somebody else’s GPT.

Stephanie Ward: Yeah.

Laurence Pratt: at that point, you’re sort of [00:32:00] reaching the limits of what. Chat GBT can do before you go up to the next level, which is something like, a platform like, n8n where you are basically, do you know, it’s connecting to all these tools and doing it, and that’s another beast.

Stephanie Ward: It is indeed. And do you know what it is like? It’s something that I am honestly very actively immersing myself in right now because the thing is that my clients, I know that what we do as virtual assistants is highly valuable, but also. If we don’t get with the program, we are gonna be out of the market within five years because AI is growing in the way it is.

The Future of AI Agents

Stephanie Ward: So I’ve got, I’m really excited for my call with Auto AI about to learn about what they’re doing because I think agent AI is gonna be the future using agents to do specific things like booking your calls or like just having one that just books calls. Or having one that just unsubscribes from spam emails or having different ones that do different things like putting things in your project management software.
That is [00:33:00] definitely the future. And I think what I’m really excited to learn is how to make these work for my clients so that they can use that in their businesses. Because the main stumbling block for people with Neurodivergence is backend admin, because we don’t get any dopamine from it.

Laurence Pratt: Yeah.

Stephanie Ward: the backend admin is boring.

But guess what? Ai, AI agents don’t have a concept of boredom because they are not human. They have one specific task where they do that and they do that, and you feed them the information so that they can do that well, they’ll learn as they go. So if you know, need to adapt things, you can. what I’m interested in is how the guardrails are set up for something like that.

So it just does that one task. ’cause obviously we know AI is quite expansive. I’m intrigued to know and understand more about. How these are boundaried, how we make sure that they are still learning, but we’re keeping it very, in a tight, sort of, in a tight shell, if you like, because we want them to learn, but we also don’t wanna have to keep going in there and modifying them because they’re going rogue.
And I think there’s a lot of potential for these things to go rogue, right? Like we’ve seen how chat BT [00:34:00] has, you know. Done some pretty interesting things in terms of acting as a psychotherapist and coaching people towards suicide. We’ve seen this happen, so if open source can do this, like the agents only need to do a specific role.

We need to make sure that if things are working autonomously and automated, that they are, they have these stops and checks and balances in place so that they don’t, ’cause the last thing you wanna be doing is neurodivergent, is having a dig at the backend admin, ’cause the backend administrator has gone wrong.

Right. So, I’m really intrigued to learn more about this. It’s very kind of new in terms of what I, what my knowledge base is. Mine has always been sort of, you know, I’m very good at the. The generative engine type stuff, but the genic stuff is still quite new in my knowledge base, and it is something that I know I need to be a huge sort of adopter of to make sure that my business is future proof because people aren’t gonna replace me with that thinking.

It’s the same thing.

Laurence Pratt: So, I mean, I think just to go back to the sort of the thing that you were just saying about, what the danger is of letting it go rogue. I think another thing as well that we are not. [00:35:00] we are not stopping to sort of think. And question is some of the information that is going into some of these custom gpt that we might be making for ourselves or people who’ve made and are selling is IP Where have they written that all themselves or are they just ripped off, somebody else’s, book or whatever. and so, there’s a lot of that going on without a lot of sort of about what this is doing for those, the people that have created those in the first place.

Stephanie Ward: Right.

Laurence Pratt: but in terms of, yeah, I, I think the exciting thing. About sort of, looking at, your own business and going back, going full circle back to where you, where we started in terms of sort of saying how lots of different platforms are having AI built into them. I think like a goodof how you might be able to look at your business as a whole and how you can integrate AI into that is to. [00:36:00] To look at the tools that you use,

Stephanie Ward: at what AI capabilities they’ve got
those good ones. Like for example, I use go high level for my CRM, and that’s got really good. Sort of tool, AI tools in it for doing outreach or, picking up conversations from in, Instagram or whatever. And then, and then it already has the workflows for, pipelines and you can upload knowledge documents into that. So if you’re thinking as your business a whole is are the knowledge documents that I need to communicate to new employee or an AI employee about how my business works, is to set those up and then think which different tools can I upload these into? go along your whole, Pipeline or customer journey and think at which points do I need, like how much can this tool do? Where do I need to go to build something in chat GPT or use something in chat BT [00:37:00] GPT. then I suppose what we’re talking about with the agent is how do you then connect all of those in an umbrella where you say, once that has done that, grab that information from here. Send it up to this knowledge base in the sky and then go and a performer task in the next tour. And I mean, it sounds simple, but it’s a lot of, it’s a lot of work. and this is the thing, it. To, in order to make all that happen, you’ve got to go back and do some very sort of human work, which is sitting down if you’re, sitting down with someone or with a pen piece of paper or a whiteboard or a digital whiteboard and say, what is a, what is the A, B, C?

Laurence Pratt: What is the journey that want something to go on before you can start. just sit down at chat gt and say, oh, can you make my business work?

Stephanie Ward: Well, no, you can’t. And the thing is as well is that [00:38:00] like it is a big piece of work to do, but it’s a very valuable piece of work to do anyway, even if you are not populating a chat bot with how to run your business. something I’ve learned and scaling in the way I have in the last five years is that if you don’t have robust systems and procedures in place, you’re setting yourself up to fail.
we’ve had to do quite a lot of reactive stuff because my business blew up overnight and it was the scariest thing, and I’m still catching up from that. But, like we had to get things like HR policies and process. You wouldn’t hire someone without having basic HR policies. So why are we deploying AI into our businesses without giving it the rules first?

that really baffles me in a lot of ways. But there is this human piece that needs to be done because we are the lifeblood behind our business. Our AI needs to be reflective of who we are and how we want to do things. And I think the best advice I would give anyone is pick a tool that you like and work with that tool almost exclusively.

Not exclusively, but almost. So like you, go high level, work with high levels in great stuff, but build on that and work on that. Use [00:39:00] chat that JPT to, you know, generate some of the stuff you feed into that, but really try not straight out with those two tools. For me, we use Clickup and I love Clickup.

I don’t love their ai. I really don’t love it. So what I do things like that is like templates and stuff. But what I do is create my knowledge base using some of my Sentra stuff and then populate that in a Wiki in Clickup. So Clickup has got a wiki on how I run my business, so that if I go to use another tool in future, I got my knowledge base captured and I can put that into a new tool.

And it’s got everything from how often I want to post on social media, what I wanna post about right down to who replenishes the first aid kit.

Laurence Pratt: One day that might be relevant to ai, but right now it’s relevant to the humans and we need to treat the AI as we would a human employee. You wouldn’t deploy an apprentice.

Stephanie Ward: It’s right, go and do social media marketing when they haven’t done any of their qualification.

Laurence Pratt: Yeah.

Stephanie Ward: So why are we getting AI with just got limited knowledges of knowledge of our business and given it tasks to do like that.

Laurence Pratt: think we need, like when it comes to programming these things, we need to start, [00:40:00] like I say, building it up like it’s a 5-year-old.

explaining it really simply so that we’ve got a good starting point to go from. It’s definitely, it’s a huge piece of work worth doing because Neurodivergence hate the admin,
Yeah.

Stephanie Ward: but in order to get to the point where we have things that can do things for us, we need to give it the brain juice first.

And that is never gonna be, there isn’t a quick hack or a tip for this ’cause there’s never gonna be a quick and easy way of doing it. But you can harness the chat bot stuff to capture some of it, to make that a little bit easier.

Laurence Pratt: I think as well, I mean, you talking about having a Wiki, in Clickup and that’s something that I think is really useful. I mean, I’ve been using notion, but I really love Clickup as well, and I think, I probably want to move across to Clickup a little bit. but maybe let’s just say, you are using chat GPT, to create, post ideas or, a content calendar. You might be, there might be two or three people in your business doing that, in the marketing department or whatever. it’s really useful. Like we were saying, you know, if you are using a custom GPT, they can all use [00:41:00] that. But what happens if you’re using it just on your computer, having well, you, you said Wiki, but what if you like had like a, here’s all the.posts that we created for Instagram, for LinkedIn, Facebook, and keep them in like a database. So, and as you build that up, you can maybe keep like a record of what. Posts went well, and what didn’t. And so when you go and say, right, I need,to write a post for this, can say, you can grab one of the successful ones and say like this.

’cause it’s always work works better if it knows what you are aiming for. then, I mean, it gives you a sense of, you know,reflecting on what is working and what isn’t as well.

Stephanie Ward: And I think, you know, you could use AI to gather that information up, but you could say to your chat bot, look at this social media profile. Based on everything that we’ve posted in last six months, what should I be posting about? You know, there’s tools that do that. what is it? clue Labs.
They’ve got a thing where it looks at your social content, looks at what’s going well, and it makes suggestions [00:42:00] for it. That’s an AI based platform that looks at your metrics In Hunt has done a great job of that. like I, there are so many tools right now that have AI built in that can do these jobs.
One of my bigger sort of worries is that we are using too many different. Things and it’s so easy to go the near as shiny object. Exactly. Because we like dopamine, we like things to be novel and exciting. So we go with these things and like you say with notion, I use notion for ages, but the problem I had is that you can really rabbit hole on that, right?

Like you can go so in depth and build it. I’ve got a, I’ve got an, if I die. Thing built in notion with my like advanced directives and stuff and all of my like financial account information. So if I died tomorrow or if I was incapacitated, you would know what to do with my end of life care. You would know what, how to cancel my credit cards, you know who my car insurance is with all that’s captured, right?
But I built all of this and then never used it again.

Laurence Pratt: Right. So what I like about Clickup is as a system is that it’s, well, it, for me, it feels like if Notion and Asana had a genetically perfect baby, [00:43:00] but what I like is that you can change what the interface looks like. So you can keep the like that novel piece going.

Stephanie Ward: So I’ve never had to change tools from that because of the fact I can create like new and exciting within that.

Laurence Pratt: Yeah.

Stephanie Ward: The problem is that there’s something new and excitements that we go with this tool and actually realize Click probably had that built in.

Laurence Pratt: Yeah.

Stephanie Ward: So I think trying to keep our tools to a minimum and focus our efforts, if we’re gonna be working with ai, focus our efforts into the tools that where we need to invest and stick to that, which is hard.

Neurodivergence don’t like that. But if we are gonna be investing time and building these AI models, we need to really be focused on where we’re doing this. We’re not wasting time.

Laurence Pratt: We are the queens of wasting time.
Yeah. I mean, and I think it goes back to, what, what was saying? about going back to piece of paper and on a whiteboard is to literally spend that time mapping out your business and what a successful business from, acquiring a client for them, onboarding them, delivering something for them, making them come back and then reaching out [00:44:00] to, to, to new,prospective leads and everything. You can map that all out and say, right. What tools do I need for that?agnostic of what you’ve just seen in your feed and what somebody said on, on LinkedIn is just say, what do I need to make this work? And then start from there. And I think if we get, we are terrible for object syndrome and I mean, I mean you need to look at my subscription bills. It’s insane.

Stephanie Ward: I, I can, I can relate to that. I went through about two months ago just to have a look because I thought, oh my God, I’ve left my boyfriend. I need to make sure I’m not wasting money ’cause I need to buy the house. And when I went through my, I was spending 35 qui on TV channels on Amazon. excuse me, like when did this happen?

Like I’ve clearly wanted to watch TV show, pay for a subscription and just forgotten about 35 quids with them. And We do this with software all the time in our businesses and we’re wasting money, we’re wasting our time. We’re dividing our efforts. If our efforts are [00:45:00] really focused and really channeled and we let our hyper focus go on just one tool that we are happy with and committed to, we can accomplish some great things.

But I think the reason people aren’t making the progress they want and make with their AI is that they’re trying to do it in too many places.

Laurence Pratt: one of my favorite things, I’ll have to say copilot. I wasn’t a fan of copilot at first, I wasn’t okay. But I have a client who works for the government and uses AI for things like grammar and accuracy check-in.

Stephanie Ward: But the problem is when you work for the government, you can’t just feed government secrets and chat pt because if you are feeding them like the energy policies for example, that’s not okay. But what’s really cool about the Intel core ultra processes is that the AI is held inside the computer so it can pull down from the internet, but it doesn’t push back out again.

Laurence Pratt: So what’s really cool about what they’ve done for is get her a laptop with the Intel core Ultra. I think she’s got an I seven, I’ve got an I nine in mind ’cause I need something that can keep up with all of this.
Yeah.

Stephanie Ward: but if she can feed anything she wants into her copilot, but it doesn’t take it back up to the internet.
So. As much as it’s pulling [00:46:00] down from, the language model, it doesn’t take her secrets back to it. 

Ensuring AI Security and Privacy

Stephanie Ward: And what we were saying before about the intellectual property, but also GDPR, I’ve seen it where people have fed their client database at chat GPT, and that’s now public information that is so scary.

Like we feeding our best ideas, our best brain source, our creative novel idea into something that other people can pull down from The first thing we need is make sure the language model that we’re using is secure enough to deal with what we want it for. Before. Before so many of us even got started on ai, we should have known a lot more.

And I think now people are being a bit more careful because they’re hearing the horror stories, they’re seeing the consequences of people screwing it up. But AI was very much a thing that appeared overnight and it was exciting and it was great and people were writing sea shanties for their enemies.
Okay. That might have been me to be clear, but I wrote Channey for a guy who was bullying me online. it was one of the funniest things in the world. but. We’ve leaned at this tool for our day-to-day operations without really knowing things like basic data [00:47:00] security. We’ve got, I think as a population, we’ve got so much to learn about ai.
The general person’s knowledge of how to use it safely is very weak,

Laurence Pratt: Yeah.

Stephanie Ward: and that’s one of the biggest take that says, make sure you know what that, what the tool that you are using. Make sure you know the rules before you start giving it anything.

Laurence Pratt: And that’s it. I mean, it’s like a balance between, I wanna be cautious about, privacy and information and everything. But also I need to keep up. So there’s a desperation to launch into it as well. So it’s, yeah, it’s exciting. It’s,Daunting at the same time, but it, I enjoy exploring.
It feels like a bit of a playground, but I think, yeah, once it settles down, you can, we can really do some interesting things. E even just like you say, like on the, from, neurodiverse point of view, just helping to say,help me break this problem down. Or, you can get like an agent that knows like your whole sort of. Life.

Stephanie Ward: it can say, yeah, that’s a [00:48:00] great idea. I can help you break that down,
you’ve also got 17 other things, so shall we park that or,
I come across Goblin Tools before.

Laurence Pratt: I’m not, don’t think so?

Stephanie Ward: Oh, goblin’s great. So one of my clients uses it quite a lot because one of her biggest problems is like sequencing steps for things, right? So you say at Goblin, I need to write a social media post, and it breaks it down at steps for you.

Laurence Pratt: Uhhuh.

Stephanie Ward: So That’s really cool.

But again, it’s a separate tool, so it’s like how do we get these things to be emulated in the platforms that we want to use? There are so many useful things for ai, but I think the reality is if we don’t keep up with what’s going on with it, our businesses are gonna be obsolete because our businesses need us to embrace the trends.
the trend became being online for your business, so everyone had to get on board with social media marketing,

Laurence Pratt: the trend was, we need to use ai. And we all just started using chat. BT like it’s our new bestie and.it’s hard to keep up with the trends, but if we don’t do it and don’t do it safely, we aren’t gonna have businesses left over to see the outcomes of it.
Yeah. [00:49:00] 

Final Thoughts and Contact Information

Laurence Pratt: Well, we’re gonna have to wrap it up there. but it is, I mean, I feel that we could carry on talking, for ages on this and it will be very exciting to, see, some time in the future where, you know, any exciting things that you come up with, it would be great to have you back. but yeah, thank you so much for, geeking out with me.

Stephanie Ward: I love it. Like I’m a biggest nerd in the world. Like I, I just, I get so excited When I had my autism assessment, I was talking about these things and I’m going like this. And I previously said to the woman, I don’t stem and I’m sitting here flapping my hands like, oh, that, yeah, nerdy makes me.

Laurence Pratt: Right. Cool. Well, thanks Steph for, joining me and, just to remind everybody where they can, reach you online.

Stephanie Ward: Yes, find me on Instagram. I am ceo Spicy Brain Collective. I’m also on Facebook as Stephanie Stevenson, ward Stevenson with a v. And yes, I know my name is Stephanie Stevenson for a while. We don’t take the mick out of me. you can also find my [00:50:00] website at the spicy brain collective.com.

Laurence Pratt: Yeah. Perfect. All right, well thank you so much and take care.

Stephanie Ward: Thank you very much.

Laurence: Well there, we have it. Thank you so much for listening this far. If you want to hear more episodes, then please subscribe on YouTube or whatever podcast platform you use. It really helps us spread the word. So if you know anyone, this episode could help, then please share it with your friends. If you want to follow me on social media, I am on Instagram at ADHD underscore goals. And you can find me on Facebook too. If you want to get into touch with the show, then you can email me at hello@adhdgoals.co And finally, if you’re struggling to manage your ADHD and you would like me to be your coach, then please head over to my website and get in touch. Until next time. Bye for now.