Introduction

I'm going to attempt to show you the ways we can prototype a solution. So just like a really quick and dirty way of crowdsourcing problems, and we're going to

go to crowdsource some problems from all of you guys. We're going to brainstorm some solutions with LIMs, and then we're going to prototype a solution on the go.

So, let's start by, if you have your phone or your device with you, if you can go to this QR code. So if you can scan this QR code, you will see this survey.

Okay, so we have results coming in. We have lots of recruitment people.

No surprise, technology, consulting, healthcare, education. Great, so we have 14 people participating.

Let's try to get up to 30. I can see there's probably 40 people in the room.

Consulting by far is the biggest. That explains why you guys use LLM so much. Consulting is probably the one industry that's adopting AI really fast.

Okay, so we have a few people are still typing. Okay, so consulting technology software.

And then let's go to the next one.

Engagement and Data Collection

What is a job role? So what do you do in your industry or at your company?

Have developers? Data scientists, CEOs, software engineers, lots of technical people, lots of management people as well.

This is great for you guys to know who is in a room. So we have a networking segment afterwards, so for you to know who you're talking to in a room.

So we have mostly founders, CEOs, and developers and software engineers, and some business management, business analysts, project manager. Okay, this is great.

Identifying Challenges

Now, last question needs a little bit more typing. What is your biggest challenge in integrating AI into your daily work? I'll give you a little bit of time to think and type.

Very curious if you guys share. If you could elaborate a little bit more on why your boss is the blocker for AI adoption, that would be great. Curious to know why.

learning how to build AI agent from scratch, sharing results, rate limits, change management, time hallucinations, I think there are lots of comments around data, the quality, privacy, hallucinations, trusting code written by AI, cost of tools, privacy, overly positive outcomes, scaling, We'll try to challenge AI a little bit today to not only get positive outcomes.

Too many things happen at the same time, not enough time to grasp everything. That's legit.

Infrastructure, complexity, system integration. You guys have a lot to say. That's great.

Common Concerns

That's all very good data for us to start on. A lot of security questions, I can see lots of concerns with integrating systems. Don't have the time, that's also quite common.

This is sort of, we do a lot of these surveys for our cohort when they sign up to our learning programs. And these sort of concerns are echoed across different regions. So whether it's in Lisbon or London or Mexico, actually people have more or less the same concerns. around how to integrate AI into their work. And the problem with the boss, that's also very global, unfortunately.

Building a Persona

Okay, so I think we have enough data point to start. So please just bear with me because this is a live demo. Things will take time. Sometimes these tools might break.

Okay, so do you, how many of you use personas for at work? Can you explain what, maybe you can explain what is a persona?

Yeah, so let's say in an interview, talk to high profile clients and you try to convince them that they need to buy your product, it's easiest to convince them through use of persona, which is the most relatable. So let's say if I'm selling to a CFO or let's say CRO, I will outline the challenges in the day in life. So that incorporates storytelling techniques, and be like, well, as a CRO, this is the challenges you might face, and this is how our solution will solve them. So personas are very useful in kind of impersonating the challenges, and yeah, convincing the customer, making the demo more relatable, let's say.

That's great. That's a great use of personas.

So you can use personas in all sorts of business use cases. So personas are a representation of your ideal customer.

You can use personas for marketing. So the persona might represent the person you want to sell it to. So it will build, it will incorporate data like their demographics, age, gender, location, job function.

You can use personas for, like she said, sales processes to overcome their potential objections.

I use personas, I came from a product background, so we use personas for user experience design, for product development. So when you build a persona, you're trying to get into their behavior, their day-to-day, so you can design things that will be actually useful and valuable to them.

So we're missing a lot of data here because normally to build a persona, you would want to talk to your customers, interview them, incorporate all those interview scripts into your persona. So we're taking a shortcut here just using a survey, but also you've given me some attitude data now, so I have something to start with.

Synthetic Personas in Problem Solving

So I'll just build a persona based on this first. And how we can use a persona in problem solving. So we can actually build this representation, this fictitious representation of an ideal person.

So say people in the room are the type of users that we want to target for our learning program. So we want to design a solution to address your challenges.

And when we build a persona, When we build a persona, we can then use this persona to, this is a synthetic persona, so this is based on synthetic, this is not a real person, so we can simulate an interview with this persona without taking the time and effort to get out of your door to talk to them.

So before you talk to your customers, before presenting a solution to them, we might have a place to start with by using this persona. Are you guys following?

Any questions? Yes, okay. I think I need to input some data point here.

So what I'll do is I'm going to use a tool. So I was using a voice to text tool that would just transcribe my response so I don't have to type.

My business is a B2B AI education platform. Our target audience are business enterprise customers.

Okay, so I will continue with the CTO tech executive persona. 1So with the persona you get a fake name, you get a role, you get their problems and their pains.

So here this persona, David, has the problems of navigating the hype versus what's real, ensuring scalable and secure implementation of AI tools, bridging the skill gap across non-technical and technical teams, and pressure from the board to do something with AI without clear direction. Fragmented AI knowledge across departments, burnout from evaluating too many unproven solutions.

I'm a CTO at a 3,000-person enterprise in the supply chain space. AI has been buzzing every leadership meeting since chat activity hit mainstream, but I'm tired of fluffy presentations and one-size-fits-all AI 101 webinars.

Using Personas for Solution Design

Okay, so what you can do now is with the synthetics persona, you can enrich this persona a lot more if you have more data from interviews, field studies, or just observations. But I'm going to just, for the sake of the demo, I'm going to start from here.

So you can use this persona to ask David some questions. So I'm going to ask David some questions for me to have something real to work on, because this problem of integrating AI into your workflow isn't concrete enough as a starting point for me to prototype a solution.

I like something around what kind of education format would deliver methods that I trust the most, or what KPIs that I can track to measure success. What are your concerns when it comes to tracking the success of AI adoption in your organisation?

So I'm going to narrow this down a little bit more, because in order to have a problem to solve, we need to pick something that's severe enough for people to want to invest time and energy into it, and something that's urgent enough for them to act on today.

Can you give me a prioritized list of your concerns by severity and urgency? The top concern is learning doesn't equate doing.

I think that's, for me, that's what I see quite a lot is that even when people learn what they enroll in a program, they aren't applying what they learned. Does that sound like a problem that you see? Yeah, I see nodding heads, yeah.

So does that sound like, for those of you who work in a more technical role, does that resonate? Give me a thumb up if that resonates.

We've got like five, so I think that's good. So what we can do now is there's also a back story.

Brainstorming Solutions

And I'm going to head over to another, brainstorm GPT and so normally I would have done this in a single thread because you can simply actually let me do this you can simply oops

You can call on a GPT in a single thread by using the at symbol. So let's say here. help me brainstorm five unique solutions for this customer problem in AI adoption in enterprise companies. These five problems needs to be unique, maximum 20% of overlap from each other and please also give me a scoring of their effort and feasibility and impact.

So I normally have done this in two separate steps. I would have done the brainstorming first and then scored them separately. Just for the sake of time, I'm going to just do that in one go.

Wait, we have there. So our problem is learning doesn't equate doing.

Sorry, this is very difficult to type on the tiny, tiny thing here. Okay, so we actually have effort, feasibility, and impact as our scoring metrics. And here are some ideas.

Live Labs model shadow co-pilot to pilot. Concept, in several courses, employees shadow real UI use cases happening in the business, and then co-pilot, then own a micro project with coaching. Real world immersion beats theory.

We also have an estimated effort, feasibility, and impact. So it's high effort, but also high impact.

You can obviously change the scoring based on your own common sense and your own experience, but this isn't a guideline to help you evaluate solutions. So you want to have a lot of solutions, and then you want to prioritize them based on what matters to you. So in this moment in time, what matters to me is effort, visibility, and impact.

Maybe what you want to prioritize is maybe effort and longevity or something else.

Evaluating Ideas

Now I'm just having a quick scroll of this list of ideas that I got. I think what looks good to me, mission-based learning, AI concierge, reverse hackathons, I want to do something that's very low in effort.

So, just-in-time AI Slack team micro-coaching layer. So based on this, I think I want to take something like number five, bite-sized AI micro-lessons directly into their daily workflows, or reverse hackathons. I think actually this mission-based learning looks very good and it looks like something we can prototype.

So I'm going to take this and I'm going to find a, just going to build a product roadmap requirement and we'll see what we can build with this, okay. You are an expert product manager specialized in learning and development software. Help me write a product requirement of only one pages long based on the below idea to address the challenge of AI adoption in enterprise teams.

So we are going to do mission-based learning. Just going to change this and then do it again. So if you guys work in tech teams, you know traditionally how long it takes to prototype. something that is good enough to get in front of people, right?

So this whole idea is that we're rapidly brainstorming, rapidly finding problems, scoring problems, and then finding a single thing that we think is worth solving and quickly prototype a solution that we can get in front of people so we don't waste time doing weeks and months of research and discovery. So I have a product requirement here. Mission-based learning with embedded metrics.

We have the objective. We have a problem statement. Enterprise team engaged with AI training but failed to apply it meaningfully.

And we have some goals and success metrics. So these are what we're going to have in the app as a goal that we're going to track. And then we have some key features.

This is a minimum viable product, so as an MVP, I think a list of five features looks good to me. Obviously, you can change that. For the sake of time today, we're not gonna go through them in detail, but if you're doing this yourself, I recommend that you go back and forth a few times to make sure that it's absolutely relevant to what you wanna build.

Prototyping with Replit

So I'm just going to change the, migrate this to Replit. So I'm getting this product requirement to Replit. I'm just gonna start building. So this is gonna take a little bit of time. It might take a few minutes because you're going to see that Replit is writing the code as we, so it's writing the front-end and the back-end code.

So Replit is one of, so we had a lot of talk about agents today. So Replit is a coding agent. So what Replit is able to do is not just to be a coding assistant, but actually build the thing from end to end for us and deploy it and even add a database. So it builds the front end and the back end.

But first, it's going to propose the plan. So it's going to propose a comprehensive plan for this project. And once I approve, it's going to start that end-to-end process on its own. It's going to design its own task.

It's going to design its own sort of micro task. Within this task, it's going to look for errors, take screenshots, and fix it by itself without my supervision. I don't even code, so I can't even help.

So now it has come back. even with some suggestions on what features to include. So I'm gonna say, Advanced Analytics Dashboard, why not? Auto-assignment submissions, AI-powered recommendations on new missions.

What I love about Repl.it is it's able to kind of think through this product requirement and then suggest features that could be valuable to me. So these features I didn't even think of before, but I'm going to add that to the list. We're going to approve.

Replit's End-to-End Process

and the plan starts. So what you're seeing here is that on the right-hand side, Rapid is doing this coding. On the left-hand side, we have the plan, so I can continue to chat to the agent here. I can say, I can add more requirements, remove features,

But basically, this live app is happening now. This is not just a mock-up. This is a live app. It will be deployed.

It will be clickable. And if you want, you can even add authentication. You can have people log in, build a dashboard. So it's not just a visual thing. It's actually the real app.

It's running really fast today, I think, because the Wi-Fi in the city is very fast. So yeah, you can more or less see that, you know, within what, just like 10 minutes, we have a solution that I can now maybe go to one of the CTOs and say validate and ask, like, do you think this address your concerns? If your concern was a legitimate one in the first place, if you did your research and your persona is right, you might be able to use this as a point to start a conversation.

So I don't know why it's so jumpy. Sorry if this is giving people nightmares. Okay.

So it's going to do its thing. When it's finished, actually let me do one more thing.

So I finished building the product and I want to write a song to celebrate it when I send it to the CTO. Can you help me write a song in the style of indie rock?

So this is just for fun while we're waiting for a replica to happen. So we're going to have this little song thingy.

We have Mission Mode. We shipped it. What a great song title.

an indie rock anthem, woke up to the hymn of a shipping lane, spirits behind us, we beat the rain, turn training wheels into real-world fire, now KPIs are climbing higher.

Pretty good, huh? I think this is the kind of thing we can now do is we let our agents do the work and we just write songs over here.

Well, I didn't even write it. So I'm gonna take this lyric.

and come over here and where was the song title mission mode we shipped it okay so um indie rock okay we're going to generate a song over here

Building the Dashboard

Now, this is our dashboard. Remember we wanted to build a dashboard that tracks the success metrics of AI, so going from AI learning to AI doing.

And so these are different missions for different teams. How does it look?

Shall we run it? Shall we hit run?

Yeah? Okay.

Meanwhile, we have a song. Woke up to the hum of a shipping lane Sprints behind us, we beat the rain turn training wheels into real world fire now kpi's are climbing higher no more theories gathering dust every lesson built on trust from desk to workflow task to team we made it This is how it's told. Reflections logged, the numbers sing. 30%, yeah we did that thing.

Let's check the other one. Woke up to the hum of a shipping lane. Sprints behind us, we beat the rain. Turned train and wheels into real world fire. now kpis are climbing higher no more theories gathering dust every lesson built on trust from desk to workflow task to team we made ai more than just a dream

All right, I think we had enough fun with the music. So the record is still running. It's going to take a little bit longer.

But you can see here that all the source files are here. If you want to go in, if you have a technical background, you can go in and change any code that you want. It's written all of the front end and the back end files.

once it's done with the deployment, it's going to become a link that you can share with people. So that's really, really valuable because when you share that link, it's a real solution that you can validate with your customers, you can validate with your team, and you just saw how long it took me to create.

Conclusion

So that is the end of my live demo. Thank you so much for being here.

So once again, we're MindStone. We run training programs. We also have a global AI community.

If you want to speak, you can apply to speak. We have lots of free learning resources on our community platform and we have training programs for individuals, for corporate teams.

We recently trained Mexico's largest retail brand and they just with some very simple Custom TPTs, they really transform their internal operational efficiency. And most importantly, everybody being on the same call and sharing solutions was really, really powerful.

So yeah, thank you so much for the sponsor today. I'll invite... I'll invite you up to speak and if you have any questions you can find me after the programme.