Subscribe to our newsletter to receive business tips, learn about new funding programmes, join upcoming events, take e-learning courses, and more.
Take the first step to successfully starting and growing your business.
Join for freePosted: Thu 13th Aug 2026
AI coding tools have made it easier than ever to build a product. But building a prototype and building a business are two very different things.
In this session, we explore why most vibe-coded apps fail to generate revenue and the four validation stages every founder needs to master before they write another line of code.
You'll walk away with a practical framework to test whether your idea actually solves a problem people will pay for.
Distinguish between four critical validation stages: Learn the difference between "people like your idea" and "people will pay for your idea," and why skipping any stage costs you months of wasted development
Build less, listen more: Discover how to balance building with customer discovery, and why feature creep is often just procrastination dressed up as productivity
Recognise the real signals of product-market fit: Identify the three signals that tell you you're heading in the right direction (spoiler: compliments aren't one of them)
Gail Cook is a founder and fractional CMO with over 30 years of commercial marketing experience across enterprises, including Ordnance Survey, Glen Dimplex, and Garmin.
She's built over 14 products using AI and no-code tools, learnt hard lessons about validation and customer discovery, and now helps non-technical founders bring ideas to life through vibe coding without losing sight of product-market fit.
She's an FCIM Fellow, Deloitte hackathon winner, and founder of KAIlie Tech Ltd.
Watch more expert webinars
Access a growing collection of expert-led webinars covering marketing, sales, finance, growth and more – ready whenever you are.
Lightly edited for clarity.
Ryan: Good afternoon, everyone, and welcome to today's Lunch and Learn. My name is Ryan, and I will be your host today.
For those of you attending Lunch and Learn for the first time, Enterprise Nation is a vibrant community platform for start-ups and small businesses. Today, I'm really pleased to introduce Gail Cook, who's a founder and fractional CMO.
In this session, Gail will explore why most vibe-coded apps fail to generate revenue, and the four validation stages every founder needs to master before they write another line of code.
As always throughout these sessions, if you've got any questions, pop them in the chat, and we'll do our best to answer them at the end. The webinar is recorded as usual, and a follow-up email will go out later today with the recording, so keep an eye out for that.
On that note, I'll hand over to you, Gail.
Gail: That's brilliant, Ryan. Thank you very much. It's great to be here today, and a massive thanks to Enterprise Nation for allowing us to host this.
So, folks, lovely to meet you all. My name is Gail, and I have worked in marketing for 30 years, and I got into tech over the last year, moving into vibe coding.
After my 30-year career, I'm working for big brands such as Ordnance Survey, Dulux, Glen Dimplex and Garmin. I'm a fellow of the Chartered Institute of Marketing, and now I'm a fractional CMO, but I'm also a solo founder.
A couple of screen grabs there show some of the products I've developed: Dopamine Delay, Meavera and CallPanion.
For those of you who don't know what a vibe coder is – I'm sure many of you on the seminar will know – it's someone who's building real products using AI tools without being a professional developer. You describe what you want, the AI writes the code, and you guide it.
One of the products I've developed is Dopamine Delay, and I want to tell you a little bit about that first, because it came from me identifying a problem in the marketplace. 77% of adults make impulse purchases, but these are driven by dark marketing tactics.
Think limitless scroll, buy-now-pay-later deals, and pop-ups saying time's running out for this offer. It's very different to the world of marketing when I started my career.
I looked around to see whether there were tools to help people with this, because in the UK alone there are 2.1 million people with ADHD who struggle with impulse spending. I found that nothing existed.
The data also shows that around one in three people regret their online purchases within 24 hours, so something needed to be done about that. That's why I built Dopamine Delay.
I won't go into detail about what it is, but it's built entirely with vibe-coding tools – a Chrome extension and an Android app, with the iOS app coming soon. It helps people with their impulse spending.
A pop-up appears and asks: is this necessary? Is this planned, or is this impulse? It gives people the opportunity to make a conscious decision, and that's one of the tools we built.
So that's great, but how do you take a good idea that solves a problem in the market and turn it into a tool or a site that people want to pay for? I'm going to talk you through the 10 steps we go through, and the lies we tell ourselves at each of those steps.
As I mentioned, you start with a real problem that you understand. Don't start with the technology – start with the frustration. Where are people having a problem?
I was an impulse spender myself, and I understood the problem – I'm a marketer, so I saw what was going on. I was using late-night scrolling to soothe myself after a stressful day.
Great AI products solve problems that the founder actually has, but that the wider market has too, because we need to look at who's going to pay for this. You can start by asking yourself: would I use this every day? Can I explain it in one sentence?
Does the solution already exist, and will people pay to solve this? One of the lies we tell ourselves at this stage is: I already know this is a good idea, and I don't need to check with anyone. So let's get stuck in and start building – and that's how you end up building something people don't need.
So you need to master the four validation stages. These are critical for a product to succeed before you even start to look at sales or advertising or any of those things.
First: do people like your idea? I had lots of that in the beginning, particularly with Dopamine Delay.
But will people pay for your idea? Those are two very different things. People are incredibly kind, and your family and friends will be lovely and want to see you succeed.
But you have to be brutal to succeed, and really push yourself, because time spent at this stage will save you lots of time and money later.
So first, does the problem actually exist? Do people really have this problem, or is it just my problem?
Second, the solution: do people like your approach? Will they use it – does it fit with how they behave, their lifestyle and the tools they already use?
Then, the willingness to pay: would they pay for this, or who will pay for this? Because it's not always the end user who is the payer.
For Dopamine Delay, the payers are debt charities, and eventually I'm hoping the banks – I'm working towards that. There are also the partners of people who see the impulse spending and the money going out of their partner's bank account each week, so they're a huge influencer too.
The end user may not be the person looking to pay for this, so you really need to understand that at the beginning, because it will have a knock-on effect on your targeting later on. And then there's revenue: are real people paying real money for this?
We'll talk about that a little further down, but most founders stop at stage two and start building, and stages three and four can cost you months and lots of money. The line we tell ourselves at this stage is: people said they loved it, and that's all the validation I need – and that's absolutely not true.
Choosing your AI coding stack is important too. I mentor a lot of start-ups, and this also applies to established SMEs: you don't need to learn to code.
What you do need to do is learn to direct AI. There are lots of tools out there – Claude Code, VS Code.
I use Supabase and then deploy through Vercel, but there are plenty of other tools out there. Things like Lovable will let you do everything within one platform, but you need to be aware of what you're tying yourself into with some of these AI stacks, and what that looks like further down the line.
For example, Claude Code will write into your code that it was developed with Claude Code, almost giving itself a credit on your product. If you ever go down the route of VC funding, investors will not want to see that Claude Code has written itself into the codebase.
It doesn't need to happen, but these are some of the things to watch out for. One really critical thing when you're choosing your coding stack is to look around.
I started with the easiest tools – Lovable was really easy for me to get started – but it changed over time, and the cost model changed. So watch out for cost creep: know what you're being charged for, per API call, per token, per user, because free tiers get expensive fast once you scale.
That's one thing I learned with another product I have, called CallPanion, which uses AI calls to elderly relatives. If you live abroad and want to keep in touch with an elderly relative, check in and see how they're doing; we use AI to give a snapshot of how that person is each day.
But you've got to limit the calls – they could run for 20 minutes if someone wanted to chat, and that becomes cost-prohibitive. It becomes a very expensive product, and a very high price that I need to pass on, so getting the balance right and understanding that at the beginning is really important.
The line we might tell ourselves at this stage is: I need to learn to code properly before I can start, I don't know enough to do this – and that's absolutely not true. I didn't know anything when I started vibe coding 16 months ago, and the tools have come on so far since then.
Number four is build an MVP fast. You'll hear this a lot: build your minimum viable product – the smallest version that solves the core problem you're trying to address.
With Dopamine Delay, I fell into that trap early on: we'd built the online extension for online shopping, and could solve that problem there. Then I thought, what about something for in-store shopping – a partner that goes around the supermarket with you and tells you when you're impulse spending? Absolutely not.
We needed to validate and prove that the first version would work before getting excited about additional ones. So think about what that minimum viable product is to test, and whether it generates the behaviour you want from your customers.
Vibe coding is perfect for this: you can describe what you want in plain English, and AI writes the code for you, so you can test and iterate. I tend to use ChatGPT to generate some bits, and Perplexity for my research.
I'll use Perplexity to sanity-check what I'm writing for Claude, or what ChatGPT comes up with, to make sure it isn't hallucinating. Then I'll use Claude Code to build out the actual stack, which gives me a bit more confidence.
I always ask, when we're building things out or researching different topics, to make sure it's giving us its source material. Where is this being sourced from – what credible sources?
I'm always looking particularly for something like Dopamine Delay or CallPanion, which are in sensitive areas – people who are neurodiverse, or within the elderly market. It's got to be credible and ethical, so make sure it's peer-reviewed material going into the research.
That gives you a bit of quality at the beginning. Once you've got your MVP built, you can test it and get feedback.
People will give you feedback, and you can feed that back into Claude Code, VS Code or ChatGPT, whichever you're using, and iterate on that feedback to get it right. The great thing is you can do that, but be careful that you're iterating around the MVP and not developing new features as you go along.
You're looking to improve what that MVP is, not necessarily add huge new features onto it. The line we tell ourselves at stage four is: I need to build just one more feature before I can show anyone.
We're often terrified – we think it's brilliant ourselves, but we're scared to show it to people because they're going to criticise it, and that's really hard. But that's the nature of it: you've got to put on your big-girl pants, as I always tell myself, and go out and show people.
Take the feedback – it's not personal, it's just information coming back to help you.
Number five is ship to real users early. When you've got that MVP, and you're happy with it, get it out there and get it tested, because in the real world there are scenarios that come up that you wouldn't foresee.
What people actually tell you in research, and what happens when it goes out into the real world, are two very different things. For me, I was building things in the Chrome Web Store and Chrome extensions, and getting an Android app onto the Play Store.
I'd never done this before, and I didn't know where it was going to go, but it was much more straightforward than I thought – easier to get those approvals through than I expected. That was with AI guiding me and making sure that when things went wrong, we were able to fix them.
That meant I could find real users with real problems at the beginning. You've got lots of assumptions at that point, and your assumptions are wrong until they're proven right.
The line we tell ourselves at stage five is: it's not ready yet, I'll launch when it's polished, it just needs to look a bit better. Get it out there anyway, get it in front of real users, and get them giving you feedback.
As you're building that feedback and starting out, all the customers you talk to are potential customers in the future. Bringing them on the journey with you, and co-designing with them, can get you some really great first users – first paying users – if you stay engaged with them.
Number six is set up analytics to learn – you can't improve what you don't measure. I use PostHog, and there are other tools out there too, like Google Analytics.
It's free for up to a million events per month, and it tells me exactly where people are using Dopamine Delay, where they drop off, and what features they're using or not using. That's always hard to hear when features aren't landing.
It also shows how often people return, and what the real user journey is versus what I'd assumed at the beginning. It's really easy to say, I'll set up analytics later; I need to focus on building now, but getting that data early really helps shape where you're going with the product.
It does take a little time to set these things up, but once you've done it, it's great to have. Your AI systems – Claude, or Codex, and so on – can link into those systems, and you can set up an agent to run your reports.
All my reports are generated on a weekly basis. On a Friday, I go in and look at my Notion dashboard, because Claude has gone off and linked into the systems, and I can sit down and review everything in one go across all my data sources.
That builds on number seven: iterate on data, not opinions. Your first versions will be wrong, and that's absolutely the point.
Get it out there and get the feedback on the data, and be careful of opinions. Everyone has an opinion, but be careful what you're asking for.
I've seen this a lot through my marketing career. It's very easy to ask people, tell me what you think about this, and they come back and say, " Well, I don't like that green you've used."
And you're thinking, I didn't want them to comment on the colour – I wasn't looking to change that. So manage the context in which you're asking people questions, and be specific about the feedback you want from them.
If you ask people for feedback, they will give it to you, because they feel obliged to – so they may give you feedback that's no use if you don't steer them the right way. Instead, say: I'd really like to know how you find the sign-up process – could you talk me through that, any problems with it, was it clear?
That way you can make changes based on what matters, rather than someone saying, I don't like the colour pink, which is very subjective.
There were lots of things I assumed customers wanted as I was building Dopamine Delay – that people wanted long delays and detailed reflections at checkout, and that more features equals more value. Actually, it wasn't like that.
People wanted a gentle nudge, a brief pause, something respectful and easy to dismiss. They did not want to be blocked or stopped – they very much wanted to retain agency.
So simplicity wins every time – more features doesn't mean more value; it really doesn't. Just because you like something, your customers may not, so be really ruthless with how you're building.
The lie we tell ourselves here is: I built this feature, people will love it, and I don't need to check with them – I love it, it's great. You absolutely do need to check, because you can waste time and money.
It comes back to the credits you're using to build the product – if people don't need or want a feature, or it's calling additional APIs, it's going to cost you more money. So you've got to be really aware of what you're building and what it's costing you.
Three signals of product-market fit. Compliments are not one of them – people saying, isn't it lovely that you're building this.
Retention: people coming back without being reminded. For those of you with a live product, are people coming back without being reminded – do they love it, do they use it again and again?
Do they come back not just because you've emailed them, but because they need it? Number two is organic growth: people telling others about it, not because you asked, but because they want to.
It's word-of-mouth that you didn't manufacture, which shows there's a real need and a community around it. For me, within the ADHD field, are people talking about it and advocating for it?
Then the great test: someone complains when it goes wrong because they depend on it. But if they're indifferent – if they can get on with their life without it – it's not really essential, and it's not bringing them value.
So those are three signals to keep an eye on. People saying it's a great idea is not a signal – people are polite. Watch what they do, not what they say.
Number eight is think about money early. A product on screen is not a commercially viable product.
With AI, you can make something instantly, in an hour. I've run sessions with female founders where we've vibe-coded products together, and that's great, but it's not commercially viable at that point.
Commercially viable means revenue covers costs – including your time – and customers keep paying. So it's about thinking through your pricing strategy: freemium versus paid, what your free tier looks like.
For Dopamine Delay, we're on a free tier at the moment, and getting people to pay means moving them through that next step, once we've made sure they've got value at the beginning. Think about funding for the future too: are you going to self-fund, hold onto it, or look to raise through angel investors or VC?
That really shapes how you're building. Are you building with that in mind, because your paperwork trail, audit trails, compliance and legal requirements are very different if you're planning to raise, compared with just bootstrapping.
There's a lot of work involved, so if you're thinking that way, getting those foundations right before you start building is really important – otherwise you can build into a trap that you then need to come back and rebuild.
I want to say something about accelerators and grants. Grants are really time-consuming to fill in, and they're really competitive.
They do give you great discipline to fill them in, and who doesn't love free money? But your chances of getting them are really slim, and it can be soul-destroying when you don't.
That's why I do consulting on the side while I'm building my business – it's a better use of my time and my skills than filling in grant applications. I did it for a while, but it's so time-consuming and competitive, plus the paperwork that comes with it.
As a solo founder, that's not a good use of my time, whereas I love consulting – I love helping other businesses, and I get a lot out of it too. I continue to learn and develop, rather than just sitting doing paperwork.
There are lots of accelerators out there, and you can fall into the trap of moving from one programme to another as a distraction from actually getting out there and talking to your customers. Don't fall into that trap.
Pick the accelerators that will work for you, that have the right connections for your business – there are a lot of generic ones out there. Pick the ones in your speciality, dedicate your time to that, and go in wholeheartedly.
One really good accelerator is better than three average ones. These are lessons learned from my own experience, and from some of the founders I've mentored.
And then there's unit economics – thinking about things like your API costs. The line we tell ourselves is: I don't really need to worry about monetisation now; right now I need more users, I'll think about funding further down the line – but that's the foundation you're building your business on.
Number nine is getting your first customers: nobody is coming to find you. The lie is "build it, and they will come" – absolutely not.
At every stage of your build, when you're going out to research and talk to people, think of them as people on your waitlist. Build that waitlist, capture interest early, and share your journey in public.
I absolutely hate doing that – I'm a behind-the-scenes person, so it's really difficult for me. But as a founder, you have to do it, because there's no one better placed to sell your product and talk about your business than you – you're the heart and soul of it.
So get comfortable with that, and get comfortable doing sales too – it goes with the territory. If you think you can get away without it, then maybe founding a business isn't right for you, particularly if you're a young, solo founder.
Engage with potential users, don't just broadcast – you've got no brand credibility or trust at the beginning. Once you launch, do the footwork; it's unglamorous.
If you've got a physical product, talk to people directly – whether that's a chemist's team, your local library if you're publishing books, or craft fairs, but go and talk to physical people.
Twenty books placed in a library, where they can sell them at the counter, or in your local community centre, can be easier to secure than 20 people online.
And online sales – it fills me with horror when I see people online, and so-called tech pros, saying they got all this traction from TikTok. I was on a seminar with TikTok recently, and they said that if you want to get real traction, they're looking for you to spend £1,000 a week on adverts.
So this is graft – this is hard work at the beginning. First sales come from hustle.
And finally, number 10 is know when to cut: build less, listen more. AI makes it really easy to add features – too easy – and that's feature creep.
In all honesty, feature creep is just procrastination dressed up as productivity – it's avoiding the things we actually need to do. Every feature you add is one more to maintain, one more to explain, and one more to support.
Things I've cut, when I fell into that trap myself, include over-complicated features nobody asked for, and trying to be on every platform at once. And perfectionism – worrying that this is your reputation, that people are going to judge you – don't worry about it.
We all get judged in our lives regardless of whether we're building a product and putting it out there or not, so go for it, you can do it. Another line we tell ourselves is: the boring stuff like legal, IP, admin, can wait till next week.
Keep on top of it as you go, because it can become really overwhelming. You're building a business, so treat it like a business, not a project, and it makes it easier.
Use AI to build agents that can handle a lot of that admin for you, but stay on top of it.
What I do differently now, in some of the learnings I've taken from this: start charging sooner. Free users are great, but I need to get paying customers first for some of my products.
Talk to users earlier – surveys are great, but one-to-one conversations are where the real insights are. Set a tighter scope from day one, because AI makes building so easy that you can lose focus.
Constraint is your friend, so write it down first, before you even start coding: what's your vision, where are you going, what are you actually building? ChatGPT and similar tools can help you with that, but keep the scope tight to get to your MVP.
Do the uncomfortable tasks first – legal, IP, pricing. You're building a business, and that foundational stuff is really important.
And ask for help. A solo founder doesn't mean being alone – there are communities like Enterprise Nation that exist for exactly this reason.
Reach out for help; other founders are always so generous with their time. Please reach out to me too; I'm always happy to have a chat with people, but don't suffer alone – don't struggle, because somebody will have done this before and will be able to help you.
So that's all from me. I'm really happy to take some questions, but there's a free guide to getting your first paying customer available on the QR code, if that's useful to you. Thanks very much, folks.
Ryan: Amazing, thank you, Gail, that was so helpful – really, really interesting. Some great feedback coming in, saying fabulous session, really helpful, brilliant session, so that's really good, thank you.
Couple of questions – I also loved the part about just getting something out there and getting feedback on it. So, Leo was asking, could you talk a little more about the discovery of a problem and the audience – how do you find out if someone has a problem?
Gail: A great way of doing it is to go into chat groups – it depends on your problem. Happy to chat about this afterwards too, Leo, if you want to ping me on LinkedIn, because there are different approaches for different audiences.
Find where your audience is, who you're targeting – things like Facebook groups, Reddit chats, wherever people are talking. People love to talk about their problems, so you can go in.
My phone is full of screen grabs of chat groups where people talk about their spending problems, whether that's related to Dopamine Delay or CallPanion – people talking about struggles with older relatives, and things like that. It's gold, and you can find users from that.
I've found quite a few people for my one-to-one interviews from those platforms. But be really careful when you're doing that – don't go in trying to sell somebody something.
Instead, go in and ask them: I hear you've got that experience, I'd love to understand more, I'm trying to build something in this space, and your views would be really valuable – could we have a chat? You're asking them, rather than saying, I've built this product, do you want to buy it?
Ryan: Yeah, that's great, really good – some really good points, and I love that about going into groups like Reddit and utilising where those communities already are. Amazing, that's brilliant.
And on your point about asking for feedback, and being really specific about what you want to know – would you say it's best to ask for feedback from someone you know, who might be a bit nicer about it, or from someone you don't know, with no personal connection?
Gail: I think we all know the answer to this one – the people you know are going to be nice to you. If you trust them and know them well enough to say, okay, I want your brutal feedback, that can work.
But we carry a lot of emotion when we're talking to people we know – they want to be nice to us, because they don't want to damage the relationship. You don't really want to hear awkward things from them.
So that feedback has a tainted lens, in my view – it's not independent. If it makes you feel uncomfortable to ask, and uncomfortable for the person you're asking, that's probably the right person to go to.
And honestly, people love to help. It comes back to that simple ask: I wonder, could you help me?
When you do it a couple of times, you become more comfortable with it. People will say no, but that's the worst that can happen – they say no to answering your questions, or no to giving feedback, and you say, that's fine, thanks very much, appreciate your time, do you know anybody who would?
It's tough – honestly, it's probably the toughest bit. If you can't get through asking people for their opinion on your product, there are other things coming down the line that are tougher again.
So build that strength of character to carry you through the tough times.
Ryan: Yeah, I think that's a really good gauge as well – if you've got that feeling in your stomach that it's a little uncomfortable, that's the time to do it. Totally, amazing.
We'll do one or two more – just a question from Rashi, saying people don't always respond to that question about getting time to talk about their problem. Would you have any templates or suggestions to help get people to respond?
Gail: What I'd say is it comes back to asking for help. In all honesty, having a one-to-one with someone – going to an event where people are, or meeting face to face and having a chat – is so much better than sending an email or a social media post.
It's tough – look at how much email you personally get on a day-to-day basis, and all the social media posts. You're cutting through noise, and it's about people's time and how they can see the value in it.
That's why, when you're going through groups, I've had the most success where people really do have the problem, they're articulating it, and they're looking for solutions at that point in time. It's a little about asking people at the right time for them.
But in all honesty, going out and building a personal relationship with people – whether that's at events – works best. I've gone to ADHD events and seminars for Dopamine Delay, and I'm able to chat to people there and build a relationship with them.
Cold emails are the worst thing you can send. Cold chats can work, but you can also follow up more than once – it's relationship-building if you're in a B2B space.
Don't just launch into someone on LinkedIn saying, I've got this product. Start to engage with them beforehand, and build it as a relationship, because that's what you want to do.
It's a bit of a dance – you're talking to them, and you say, that's a really interesting problem. If they're talking about it in the chat room, engage with them on that problem there, and you can find out more.
Research doesn't have to mean sending a questionnaire and asking a set of questions – it's human feedback.
Ryan: Yeah, that's such a good point – it doesn't have to just be a questionnaire tick-box, does it? Brilliant.
And then we'll do one more to wrap up – a quick question from the community: what's your favourite AI agent?
Gail: Oh, no, that's it, oh gosh – my agents are probably going to be listening to me right now. They all have their own strengths.
I love Claude Code, and I love Cloud Design – if anyone's using it, it's amazing for setting up your brand look and feel, and that's been a bit of a game-changer. But I am a bit of a Claude girl at the moment.
I'm getting into Codex, and using Gemini too – there are so many out there, and they're all evolving. I'm not a hundred per cent brand-loyal, coming from a marketer, but that's the great thing – there's so much out there to have a play with.
But don't get distracted by all the tools out there – pick one and stick to it, and get used to it, because otherwise you can lose track of things. I've had it myself – where did I ask that, is it in ChatGPT or in Claude, where's it gone – you can confuse yourself across too many tools.
Ryan: Brilliant, thank you, Gail – that has been a fantastic session, and thank you for coming in. I think everyone's really enjoyed that, so a big thank you to everyone for joining.
And a big thank you to Gail – please do connect with her afterwards. I've popped your LinkedIn and Enterprise Nation profile in the chat, Gail, and there'll be an email later this afternoon with the recording.
Gail: Lovely, thank you so much, Ryan. It's been great, and thanks everyone for joining.
Ryan: Thanks, everyone. Thanks, Gail. Bye for now.
Unlock more on-demand sessions designed to help you sharpen your skills, grow your business and stay one step ahead. Find more Lunch and Learn webinars
