Janey Treleaven on AI for SMEs: Start with What You Have

Jeremy Yang: The rule that I tell people, because you've probably heard the same before, if you're not paying for a product, you are the product. I mean, advertising, 100%. But the thing with all of these like free tools out there is that they're not, you're not being advertised to. So it's not obvious.

So what are they getting for it? Nothing is free. Yeah. Is there a free thing?

You must be getting something in return. Yeah. And so I think one of the biggest risks at the moment to organizations is those note taker AI bots that come in the meetings with you. Read AI, Fireflies.

Yeah, think about that. Yeah. Because what they're getting is they're getting the video. They're getting an image of you that they can use to clone.

They're getting voice samples for you. They could be looking in the background behind your video and going, ⁓ can I see any passwords and things? No, no, no. Hey, guys.

Jeremy, the online ads guy, Digital Glass Marketing. I had the pleasure of interviewing Janey earlier today. What a great interview that was. So she's one of the OGs in the AI implementation game started two years ago before everybody started she caused the crystal ball.

You hear at the start of interview she was I went in thinking that I was gonna talk to her about how small business have no chance in AI shouldn't implement it, you know, because she's grown a lot more since the last time we spoke which was about 12-18 months ago when we're still talking about implementation. She has grown a lot and I thought that she's moved away from small business. It's actually she's in complete disagreement with me on that and that's where we started off.

I think she said no, there is a lot of things that small business can do. Her philosophy is very different to most of the implementers out there and you hear all the stuff that she used to do and the grind to get to where she is. and what she thinks the future is going to be. So it's a really good interview.

Check it out guys and let me know what you think at end. Cheers. Welcome back to another episode of Show'em with Janie. Janie, how are you?

I'm good thanks Jeremy. Good, good. She's like the OG of this AI implementation, execution type of things. Right.

And we spoke about, think even when you first started, maybe about two years ago now. ⁓ Yeah. How has things been? Things have taken off.

My crystal ball that I had two years ago is definitely, it was working. It was working really well. ⁓ Cause yeah, when everybody else, know, 2023, chat, GBT first release. Yeah.

Wow. This could add a lot of value. For me, it was just this lights on. I need to do something with this.

And, and so I actually sold my apartment. to fund my business. really? Yeah, I all in on that.

I look back and I'm like, what was I thinking? I think what we talked about when we were talking when you were doing the discovery session with my thing. Yeah, really that apartment with the boat, not the boat, the ⁓ rowing thing. Yes.

You know, I thought it was my forever home. Yeah, but this is my forever business. Yeah, starting intelligence assist was the best thing I could have done and believing in my crystal ball. Yeah, it was telling me, yeah, hey, AI is going to, you know, remember in the early days, everything you asked AI to generate it always a game changer.

Yeah, game changer, potential unlocking. Yes. Yeah, some of those phrases back in the day. Yeah, thank God that's gone away.

Okay. What about the dashes? The long dashes? Are they still there?

Some of still there, yeah. Yes, there's pesky pesky long dashes. Even Sam Altman actually apologized in this post-pre-Christmas last year. It was from Chachiputte's account.

We apologize to the world for ruining the long dash. Yeah, well don't even know how to do it on keyboard, but you're going back to that. So now, as I'm still on a networking scene, like every week. Every week there's people coming in, we want to help small business with AI, right?

⁓ This is what I think, and I love your point of view as well, even if it doesn't agree with me. I don't think small businesses, small businesses, we're talking about a plumber with three vans and a small accounting firm, I don't think they can afford it to the implementation, that's what I think. And what I mean by that is I don't think they can afford getting it wrong. So they're waiting for everything to converge.

They go, all right, it's converged. It's landed on this thing. Now we're going to do it and we're to do it ourselves. If it's more business.

Love to hear your view as well. Look, I understand why businesses are feeling that way, but I actually totally disagree. Okay. Because you don't have to go and build something AI.

It's kind of like going or does my business need email? Yep. So I need, you know, finance system. Yeah.

AI is just the latest thing that you do have to add. And of course, it does come with a cost. But it can be done very affordably. Yes, by using the AI in the tools that you already have.

So most businesses have chosen just like people choose am I going to be Android or iOS. businesses have chosen am I going to be Microsoft or Google, right? The workspace. That is where businesses need to start with AI adoption is using the AI capabilities, what I call a core AI tool.

So it's the day to day help me with something AI that starts to get them on that journey of, you know, doing everything themselves, to doing it with AI, to eventually, and this is where it does start to get expensive, is when you go to the do it for me AI, the agentic AI. But you shouldn't start there. You've got to get the foundations right first. Okay?

So you need to start by going, what is our core day-to-day AI tool? And you choose that by considering three things, the capabilities that it has. the security that you need for the type of data that you handle, but also the knowledge. Because AI, when it relies on its trained knowledge, that's when it gets things wrong.

It can't, you know, every AI model, engine, large language model, whatever you want to call it, they all have a bias towards the people who've trained them and the data that they've chosen to train them on. And that's why we see a lot of the American models have that bias initially towards you know, long dashes, quite a lot more verbose than yes, American professional is very different to Australian professional, right? Yes. And so they all have a little bit of a bias in and it's also been trained on data that may not necessarily be verified or accurate.

Okay, so you need to make sure that the AI that you're using has access to the knowledge that it needs to be able to do the task. and then you layer the instructions in over the top of it. So let's say you're using AI to help you write an email. It needs to know the context of the email that you're applying to.

But it also needs to understand what your tone and style is for when it's writing that email. If it has access to your sent emails or your recent emails, it's going to be able to use that. to write in the way that you write. Okay, so that's why it's important to think about where is the knowledge that the AI needs to access?

The knowledge is already probably in, and most businesses think that their data is in all of their SaaS systems, like their Zoho's or CRM's or other things. No, most of their knowledge is actually in their emails and in their drives. You know, all the Word documents and Excel spreadsheets, that is 70%. They say 70%.

of most companies knowledge is actually in those. So where are they getting a role? Where is the misconception when it comes to the small business and thinking it's going to be very expensive? So basically what you're saying is find a tool.

There's only going to be two or three main ones, right? You land on. And so based on your experience, it's going to be quite obvious as soon as you talk to someone. Where it ends?

first question I often ask is do you use Microsoft or Google? If you're in Google Workspace, then the answer is probably Gemini. If you're in Microsoft, then it's probably Copilot. What do you want to use it for?

Because the answer may actually be Chachiputte or Claude or Perplexity, depending on what it is that you do. And then of course, that security side is really important, right? Because the minute you start moving data around, that's when you open up to the risk of people either stealing your data for malicious reasons or trying to, you know, data to train on. ⁓ And so that's why, you know, Microsoft Copilot, it inherits the same permissions that you've got set up.

You've got that enterprise grade security in there. So even though Claude definitely way more capable than Copilot, Copilot has better security because your data is remaining inside your enterprise environment. So if just say there's a franchise that's got 22 stores and they go, want everything to remain inside, I want competitors to touch it, then they can't be playing around on certain elements. No.

Legit, no. No. And well, you can. You can to a certain extent.

If your data, if the type of data you handle is heavy on the PII, so private, privacy policy, privacy principles, You need to be very careful with that. You might be in breach of some regulations. Companies that do a lot of work with the government have to, data has to stay residing in Australia. Yes, got you.

So like, Claude and Chachi Pati, they don't have options for hosting your data. Well, the data stays here, but the inference that it's doing is happening overseas. So they need to be careful about those sorts of things as well. But as long as you, like the rule that I tell people, because you've probably heard the same before, if you're not paying for a product, Yes, I mean advertising 100 % yeah, but the thing with all of these like free tools out there is that they're not you're not being advertised to so it's not obvious What are they getting for it?

Nothing is free. Yeah, he's ever free. They must be getting something in return Yeah, and so I think one of the biggest risks at the moment to organizations is those note-taker AI bots Yes, oh, great. I fly flies.

Do you think about that? Because what they're getting is they're getting the video. They're getting an image of you that they can use to clone. They're getting voice samples for you.

They could be looking in the background behind your video and going, ⁓ can I see any passwords and things? ⁓ No, no, no. Free is not okay. So that is the first thing I tell every business that I engage with.

Free is not, there's too much of a risk. And unless you just even pay a dollar, there's no law to protect you either. Because under like contract law, common law of Australia, you know, as long as you paid a dollar for something, ⁓ there is an obligation on them to provide you with... Got you.

So it's interesting that because I come in, because I've looked on your website, and I know that all the great work that you're currently doing with larger businesses too, right? And we came out from the BX circuit, then we see each other in BNI and I... interviewed Chris Mulati a long, long ago and we all know Chris Mulati. I thought coming here you'd be like, yep, small businesses, no, waste of time.

Don't do AI small business. So there is demand. is a lot of demand. So people are not getting a ride then.

Yeah, yesterday I did a session with a company out there branding. They do merchandise and small businesses with exhibitions. They are two people. Two people, that's it.

Their whole business and they're working as major clients. and where they need to spend. it's all about identifying. If you get time back, what are you going to do with that time?

Are you going to, do you want to go and spend more time talking to customers? ⁓ So let's work out the things that we can delegate to AI. No matter the size of your business or what you do, there's always stuff you don't want to be doing. There's always the mundane.

That's the stuff you start going, how do I get that to AI to either do for me or use AI to do it with me? And it's done faster. Yeah, so when we were talking two years ago and that was at the prime when like spreadsheets wasn't even very good. You can't even upload stuff, right?

So one thing in my business was always analytics. Sitting around analytics, I still find to this day that AI is not very good at joining the dots. When I'm like, if I'm in the account, I'm looking. I can see stuff.

go, well, AI is not good at it. then I say, I'd love to hear your thoughts. And I got a follow up question to that. So two of the key terms with AI is large language model and natural language processing.

Both of those contain the word language. AI is good for the qualitative, not the quantitative. Not yet. Not yet.

Well, how they started to solve for that by giving AI the ability to write code and execute code. So it's like having a data scientist in your pocket. So you ask the question and then you got to make sure that it's in the right mode to then write the code, execute the code to give you those insights and things like that. all right, cool.

right. So that's one of them. ⁓ And I think that I lost my train of thought now. yeah, so okay, that's what it is.

Yeah. And so it's about making sure that you're using the right AI tool in the right mode so for anybody listening who's Microsoft yeah okay and you've got copilot if you have copilot premium turned on so not the basic as basic is a bit basic yeah except in outlook it's very good now look yeah if you're in the bay in the premium version in the Microsoft 365 copilot app there is an agent in there called analyst If you go to Analyst first, you don't start just a generic chat with Copilot and then realize you want it to do some analysis work.

You got to go to the Analyst agent. it will already have the right capability to be able to do that coding and answer the questions and even generate new spreadsheets and with pretty good accuracy as well because it's writing Python. It's writing a sequel. It's executing on those.

It's not relying on its knowledge or trying to people please. Got you. Yeah. Okay.

Got you. So it goes into a different mode on its own and then it kind of does it. Yeah. So one of the experiences I remember I was going to ask you now, so one of the experiences, I don't know how much are you getting this, is like when we send reports out and send quotes out, the clients now run us through AI.

They don't read it anymore. And they'd be like, oh, I'll spit out five recommendations from this. That's actionable. the recommendations are very bad.

So you go, what's your account? Do you run into that a lot? So one of the main things AI is used for is to summarize. Yes.

Okay. So like it's always in your email system. So whether you're in Gmail or Outlook, there's always a, would you like me to summarize this for you? How does it know what's important for you?

Yeah, to summarize. It doesn't have that context. So it will likely miss something. And so it's all about making sure that you layer in enough instructions.

What do you want the summary for? What do you want those recommendations for? Yes. And what is the context with it within which it's recommending?

It may not even know who your business is. And so this is why allowing AI models to rely on its trained knowledge is really not a good idea in business. Yeah, you have to layer in the right instructions. Yes access to the right knowledge in order to be able to do a decent job for you Okay, I'm gonna ask you a question.

Yeah. Do you think the earth is flat? No, do you think one times one is equal to two? No, okay Did you know that there are a lot of people out there that believe both of those things and they've written a lot of content Yeah, and a lot of that content has been used by the various different companies to train their models Yeah, and so in the models training.

Yeah It believes that there is a possibility that the earth is flat and that 1 times 1 is equal to 2. Yes. Okay. So in the moment when the AI is doing the task that you asked it to do and it's doing, okay, based on my training and my ability to reason and infer from what you're asking, I think this time I might use 1 times 1 equals 2.

they might just go that, yeah, okay. Yeah. Okay. And so that's why you have to...

Control the AI you have to ground it. Yes knowledge you want it to Yeah, use that's really good verified grounded. Yeah And you need to Allow it to rag which is retrieval augmented generation. So retrieve before it answers.

Okay, so there you're to control mechanisms I'm gonna ground you to justice knowledge. Yes, you have to answer just from this. Yeah, or I'm gonna allow you to go and ask a friend going back to the World Wide Web or go into my files and... Retrieve before I generate.

Yeah, but you can't just say go and look at the world wide web. It's full of fake news, right? you need to control it by saying rag only to these sites that are verified So if you're asking it for like an opinion for something for the ATO, yes, make sure it's looking at the ATO website Don't just ask it to go out and tell you you forgot to give it the instruction you got to treat it like an intern Yeah, and you've got to be really really clear with it about what the rules are.

Yeah. And we call those guardrails. Yeah. Making sure that the AI doesn't wander in terms of...

Guardrails simplify everything you just explained. It's one word. But it's not though. The people need to know that deep, right?

Because people got guardrails, okay, guardrails, and they just kind of move past that. I've never had explained the way you're explaining it. Do you what I So, no, that's really good. And I will also say that with all the people that I've spoken to about AI...

It's going to sound bad, like a lot of it is they are more of a front for the thing. But when I worked with you, you know, the discovery session that you helped me with and all those icons. Yes. Right.

That's real. You know, that takes real R and D to get to there. That takes like hands on. And you had alarms at 2 a.

m. in the morning because you were gonna like they were releasing some beta and you're like, I'm going to be part of the beta. And that's crazy though. You know what mean?

That's real on the ground stuff. And that's what people need. is when I set up Intelligence Assist, the first thing I said to myself is I'm not building software and I'm not selling software. I'm selling AI adoption.

How do businesses adopt AI? Every business has, the same as we all have for all of the streaming services, subscription fatigue. Software as a service came with a lot of subscriptions. It's great, it's helped, but ⁓ gosh, we don't want to be adding more and more software.

And so a lot of businesses out there trying to do what I do. of intelligence, they are a front for software. They're trying to sell you their AI software. You don't need new software, you need to use the AI.

in your software that you've already got. Maximize it. Maximize your investment in the tech stack already because AI is more about the people and the process around the tech than it is about the actual tech. So you've got to come at the AI opportunities.

So that blueprint discovery that we did, that was about identifying all of the things that you do across your business and what things could. be augmented or innovated by giving AI, you know. And it's very unique, I think. I think to do that, think people will like guys might see a war of icons and get excited.

It's like a game, you know, it's like skill sets, right? But business owners who got real budgets, real pushback from people who don't want to use that, they're going to want to know which ones you use. They go, yeah. So in your implementation with small businesses, I guess you have to be, ⁓ different horses for courses, right?

So small business, different spiel, medium business, different spiel, training larger businesses who has teams who don't really want to adopt different spiel again. Every business, regardless of its size, is pretty similar. They all have front customer-facing functions. They find customers, they engage with their customers, maintain those relationships.

They have a core product or service that they deliver, and then they have a back office where they do all the business-sustaining functions. Every business, regardless of size, in a small business, might just be two people doing everything with only a couple of software pieces. Or in a larger business, it's just going to be layers and layers of the same thing. Before I started up Intelligence Assist, I worked at Commonwealth Bank.

That was my history. ⁓ They're not that dissimilar. ⁓ Every business, regardless of its size, needs to make that Microsoft versus Google, iOS versus Android decision. Yeah.

QCore AI tool. Yeah. And then train people. I call this the triangle of successful AI adoption.

You need to have the balance of the value, which is the tools. What is the AI actually going to do? You need to have the controls in place to make sure that people know what they're allowed to do. And then you've got to have the literacy.

You have to teach people. AI is a great leveler. I've seen some very old elderly people really embrace it. And young people and everybody's on the journey.

People are just at different points. And so the hardest thing for most organizations is that variety of people who are being who are. really against it and a little bit fearful. And the people who are just waiting to be told they don't want to do the right thing.

They're the people who are experimenting, have experimented and didn't really think it was that great. Yeah. Yeah. Which is okay.

That's okay. And then there are the people who are really proficient. So when I go out and do literacy training courses, I have to try and like target the whole spectrum. Yes.

That's the thing that for me, right, because I never worked in a big organization. And when you sat across me and we did discovery and all that and we're talking about the Commonwealth, the teams you started, India? Yeah, so you were starting all the teams. I said, man, like the amount of people skill, I'm not trying to, know, G you up or whatever.

Like that's deep because there's people wanting to do it. People think it's going to replace their job. People think, well, I'm being used to train it. Then I'm going to get a letter to, know, to move on or something.

And like, isn't there a lot of thinking behind that before you go into those meetings? There is absolutely. And look, because I've done enough of it now, I've developed a few techniques. first thing I want to say to that is that AI is not a technology led project.

It needs to come from process and people. That's how you have to tackle it. The technology is secondary to what is the AI going to do. And it's a new team member.

and you're going to give it work to do and you're going to have to train it and you're have to give it feedback and continue to invest in it. And so that's why my skill set is actually really quite aligned to AI and AI adoption. Because even though I do have a tech degree, don't tell my team. was more on the way of working something in my corporate career.

second thing is around that fear. I have seen a big transition. People have realized that AI is good, it's not that good. So they stopped worrying about replacing them.

What I find people more worried about these days is that it's going to take away from them the things that they like doing. Can you give example? Well, some people like writing. Some people enjoy the writing process and the writing letters.

Some people like doing the analytics the old way. Design and stuff. coding. I tell those people there's got to be something interesting.

That's where you start. Start with the thing that you know how to do. If you don't know how to do something, you can't get AI to do it because you have to verify the results. have to keep the human at the core of everything it does.

If you are not a lawyer and you try and get AI to be your lawyer, you can't verify it because you don't know what your should know. So start with the thing that you do know how to do and also the thing you don't enjoy doing. and then you can just it often it is just that getting over that first hurdle yeah something I really hate that I hear a lot is people using that almost threat of ⁓ well if you don't get on board with AI you're gonna get replaced That's what's going to cause you to get replaced, not the AI itself.

I actually think it's the people who use AI badly that are going to get replaced first. Yeah, right. And so if you don't verify its results and make sure that it hasn't hallucinated, that you have grounded it properly. ⁓ You've used it carefully.

Your hand ins don't require double checking and I'm not picking little mistakes out from there. Yeah. And you know... Part of what I think holds people back from doing that verification is what I call what's known as AI slop.

Yes, yes. Yeah. The low quality, repetitive content. It's exhausting reading all of that.

tell people don't. The problem is it's not the AI, it's you. If you say to write this for me and don't say I just want five bullet points, you specify what you want in the prompt, then it is going to go over the top and do too much. Yeah.

Got you. Cool. No, it's amazing. One of the final things on my list of things to talk about, you know how you said you don't like the phrase human in the loop, you're always a human in the core?

What other cliches and stuff, what other beyond the surface stuff that have you, do you come up with? That's not like, know, it's too, like what everyone else is saying kind of thing, yeah. lay the AI foundation first. Don't go and try and...

build a rag system that's going to do everything for you. By the time you do that, the technology has moved on. Get the foundations right first. Get the core AI tool in, get everybody using it, get that productivity boost, but be patient with the boost.

Productivity is going to come in time when people start to really use it to make it sticky and go from the way most people are using AI at the moment is they'll do 15 % like, oh, I'll start writing an email. Oh no, maybe AI can write this better for me, copy and paste it over. AI does 15%, you go backwards and forwards 5 % at a time till you enter. Yeah, okay, it's kind of like an answer.

Yeah, that sounds about right. Yeah. And so, yeah. But what we need to get to, literacy in AI will actually get you to the next step, which is where you only have to do the 5%.

It's because you go, oh, I need to do something, I'm gonna go straight to the AI tool. And because in the AI tool, it's already got agents or Gems or GPTs or whatever it is that does and knows how to do that task for you. You only have to do the 5%. It does the 80 % first time.

And then you do the final 15 % for you at the end. That's the productivity boost we're looking for, right? OK. Then once you actually get to that 58015, then you can get to the Do It For Me, which is where the AI is let in and is monitoring for an email arriving and will write the draft for you.

It sees that a new order has come through into your CRM, so it kicks off a whole bunch of other tasks and it's all adventure of it. So in your experience, and you're probably the most experienced person I've spoken to about this, and that I actually trust. Okay. Have you seen it work like that?

Like perfectly? Cause I see the stuff on LinkedIn, right? They go, I've got 15 agents working for me, but I don't know if it's true. don't know if they're sending a course.

we do an intelligence assist practice, what we preach. Yes. You know, we are punching well above our weight with our clients that we're working with the pipeline of work that we've got coming in. part of that is because we are using AI.

Every day we've got agents. Every one of my emails is read by an AI agent and it drafts it for me. don't have an assistant. All right.

I like that. AI to do everything. we always, every time we challenge ourselves, could we have used AI to do that better and practice what we preach. Wow.

Okay. You know, I love this chat so much. should catch up like six months or something to see where everything is. for those business who What would be the ideal client, like your passion work right now?

Is it training or is it just like, is there like, you know what I mean? Is it just like adopting? It's coming in and looking under the hood. That's my favorite part is being led in to do a blueprint discovery workshop.

We always kick off with a three to four hour workshop where I call it peeling away the onion. Looking at the tech, the process around the tech, the people that run the process around the tech, it's a two-way sort of workshop where I get to teach you a few things that you didn't know before, try and get everybody in the room just a little bit more literate, you know, excited, excited. I see every person who's like, Yeah, yeah, and then I love finding all the opportunities and I love presenting that back to them a couple weeks later Yes, they're like, ⁓ my gosh.

Yes, so cool. Yeah, but you can't do everything. Yeah, it's a journey Yes, that is the biggest piece of advice to everybody. Yeah, it's a roadmap.

It's a journey You got to start somewhere. Yeah, it doesn't there's no right or wrong Okay, you got to start somewhere and figure it out as you go and yeah, every business is unique and so Where I'm always surprised is I think I know what I'm going to recommend they do first. What they choose to do first is often quite different. But it's never wrong.

It's never wrong. It's just what's right for them right now because their future strategy might be in play or they might have had a resignation in the past. Some secret context relationship, someone's cousin's doing something for them. There's some hidden, yeah, yeah, I got you, I got you.

That's why it's, that's my favorite part is like looking under the hood and going, okay, how does this business work? And how can I find the right stuff for forum? Yeah. And so on a small business, guess the business owner will be the one contact you on a bigger business, like medium size with a CMO, no, no, it be CMO, COO.

COO. CTO. The operations side. Operation person goes, yeah.

people in culture. It's amazing the different people who. who approach us. Yeah, amazing.

Can you just tell the people again how they find you and what's the best way to get a hold of you? The best way is just to jump onto our website, intelligenceassist.com.au and book a discovery call with me.

⁓ And I always take those discovery calls and look, you'll probably learn something in that half an hour. often find at least one or two opportunities. in that first initial call. no worries.

Just tell them the company name and the website again. I'll link it everywhere. Again, as well. Awesome.

Thank you. Intelligence Assist. Think IA for AI. Yeah, I love it.

And just through our website. Yeah, cool. Thanks again for coming on the show. Thank you very much.

Amazing.