“Most startups don’t have an AI problem. They have a distribution problem.”
It’s easy to get caught up in the excitement around AI, I think most of us got a little lost in it, at least at first. When a new technology promises to make businesses faster, more efficient and better equipped to tackle their biggest challenges, it’s hardly surprising that companies want to get involved and reap the rewards of this. But, what happens when we’re so focused on finding an AI solution that we stop asking whether AI is actually the answer at all?
Sometimes, the problem isn’t the technology at all; it’s the way a business distributes its products, manages its logistics or connects the different parts of its operations. These are problems that tend to require some really thoughtful solutions, including better processes, clearer communication or a fundamental rethink of how things work. Instead, so many companies have fallen into the trap of treating AI as plaster or a piece of duct tape, hoping it’ll compensate for problems that technology alone can’t fix.
And while AI can undoubtedly solve some very real problems (and it has already), it can also become a convenient distraction from the less exciting, more difficult work of fixing what’s already broken. More technology doesn’t necessarily mean better outcomes, particularly when the underlying systems aren’t fit for purpose.
That’s the thinking behind this edition of Quite Contrary, TechRound’s series dedicated to hot takes, unpopular opinions and perspectives that might make people shift uncomfortably in their seats. Each edition starts with one deliberately contrarian viewpoint before opening the floor to the wider TechRound community to challenge it, defend it or offer a different perspective. And this week, Daniel Preston has us talking about the fact that most startups don’t actually have an AI problem, although that may appear to be the case. In reality, they have a distribution problem that they’ve tried to solve by throwing AI at it.
This Week’s Contrarian
Daniel Preston is this week’s contrarian. As the Founder at LiveInCare USA, Preston makes the bold assertion that contrary to seemingly popular belief, most startups don’t actually have the AI problem that they think they do. Rather, Daniel tells us, they ahve a distribution problem masked as an AI issue.
So, is Preston pointing out an uncomfortable truth that nobody wants to talk about, or is he oversimplifying a much more complicated problem?
Daniel Preston, Founder of LiveInCare USA
Daniel Preston is the founder of LiveInCare USA, a platform that connects families with live-in and in-home caregivers across the United States.
Through this work, Preston has gained direct insight into how families search for care, what challenges they face when bringing someone into their home and how caregivers structure their daily routines, schedules, and responsibilities.
His focus is on making the process of finding care more transparent, flexible and accessible, without relying on traditional agency models.
Daniel’s Hot Take
“Most startups don’t have an AI problem. They have a distribution problem.”
“I’m seeing more founders rush to add AI features because it feels like the market expects it. The result is dozens of products that look increasingly similar while the harder problem, getting customers, barely changes.
“Building LiveInCare USA has reinforced that for me. We operate in a market where trust matters just as much as technology. AI helps us write better content, organize information, and automate repetitive work, but none of that creates demand on its own. Families still need confidence that they’re choosing the right caregiver, and caregivers still need confidence that the platform is worth joining.
“If nobody knows your product exists, adding another AI capability rarely changes the outcome. Distribution, reputation, and solving a real customer problem are still the biggest competitive advantages. AI can improve those things, but it doesn’t replace them.
“I think we’re entering a period where companies with ordinary technology and exceptional distribution will outperform companies with exceptional technology that nobody discovers.”
What Does the TechRound Community Think?
- Harvey Dhillon: Founder and CEO of Zmartly
- Luke Budka: AI Director at Definition
- Charles Radclyffe: CEO at EA Global AI
- Fiona Harrold: Media Strategist, Communications Expert, Bestselling Author and Speaker
- Clare Shutt: Head of Marketing for Department
- Rebecca Berry: Founder of Berrykin Marketing
- Jakub Lenski: Co-Founder of Saber
- Fiorela Imerai: SEO Account Director at Wildcat Digital
- Fernando Pedraz: CEO at Reveni
- Pavlo Kharmanskyi: Founder of OnlyMonster
Harvey Dhillon, Founder and CEO of Zmartly
“I agree AI is labour, not strategy, and it won’t rescue a founder who can’t read their own numbers. Where I part company is the word ‘usually’. The way I put it, AI doesn’t raise the ceiling; it raises the floor. Someone who asks it why, not just what, learns faster than they would from a textbook.
“The danger is the person who takes the answer and skips the reasoning. In accountancy, that has a price. An accountant stands behind the figure, and the client is still legally responsible for their own tax. A confident wrong number is just a wrong number, delivered faster.
“So I wouldn’t judge a startup by how much AI it uses. I’d ask who signs off on the number, and whether they would notice if it were wrong.”
If your team removed AI tomorrow, would your core decisions actually get worse, or just slower?
“Slower, not worse. The judgement on a client’s numbers was never the machine’s to make.”
Luke Budka, AI Director at Definition
“I’d push back on ‘AI rarely fixes weak judgment’. If you are genuinely tightening review steps and removing vague decision-making with good evaluation of whatever AI is working on for you, then it can very capably catch and correct weak judgement at scale. But someone still has to design the evals, so the thinking problem moves higher up the workflow.
“Evals are a real mechanism for catching and correcting bad judgment and something most companies, big and small, don’t spend enough time on. A well-built eval loop can be more consistent than a human reviewer who is tired or biased or rushing.”
If your team removed AI tomorrow, would your core decisions actually get worse, or just slower?
“Cutting access to a work assistant who has PhD level intelligence across multiple domains, will only ever result in less informed decision making.”
Charles Radclyffe, CEO at EA Global AI
“AI is very good at producing more stuff: more words, more slides, more dashboards, more pretend certainty. But if the underlying question is lazy, the strategy is vague, or nobody is willing to be accountable for the answer, then all AI does is industrialise the nonsense.
“I see this a lot: founders using AI to accelerate work which probably should not exist in the first place. It is swivel-chair activity with a better user interface. The real discipline is deciding what should remain human: judgement, taste, responsibility, relationship management, and the awkward bit where someone has to say, ‘I think this is wrong’. Removing AI would make most core decisions slower, not worse, unless the startup had already mistaken machine output for human judgement.”
If your team removed AI tomorrow, would your core decisions actually get worse, or just slower?
“Removing AI would make core decisions slower, not worse, unless the startup had already mistaken machine output for human judgement.”
Fiona Harrold, Media Strategist, Communications Expert, Bestselling Author and Speaker
“AI is a useful servant but a bad master. You cannot rely on it for strategy but you can use it to save you time on monotonous tasks. It would be a huge mistake to think it could replace human intuition and creativity. Getting strategy clear in the first instance is vital to keep your own human instinct firing and to avoid getting lazy, or even losing confidence in your own abilities.
“Keep AI firmly in check so you avoid obvious word salad, which instantly undermines your credibility. Use it to save time on administrative tasks such as contracts – even then, double check.
“Use AI to free up your time for more human interactions, conversations and building connections.”
If your team removed AI tomorrow, would your core decisions actually get worse, or just slower?
“If my team removed AI tomorrow, our core decisions of strategy for our clients, would stay exactly the same – we would still identify the best media route and resources.
“We’d use AI to fill out any gaps in our press list, but often we’re more up to date than AI with who and where!”
Clare Shutt, Head of Marketing for Department
“AI is most useful when employees know what they need it to do and how they can judge the value of its outputs. For smaller, scaling businesses, that judgement is especially important because
there is typically less time and budget to spend correcting avoidable errors.
“In my experience, AI can be useful in supporting repetitive tasks, like desk research and administration, freeing up more time for creative thinking and building the relationships that help businesses grow. That said, it shouldn’t replace the experience or
judgement needed to decide which ideas are right for a brand or audience, and coming together in person to collaborate remains central to that.
“I see this at our Department workspaces, bringing together startups, growing businesses and creative professionals through a shared business community. The strongest ideas often emerge through networking and those conversations in spaces which encourage community
and creative ideas.”
If your team removed AI tomorrow, would your core decisions actually get worse, or just slower?
“AI can speed up research and repetitive tasks, but the most effective judgment and creative thinking behind Department’s campaigns and activations come from people.”
Rebecca Berry, Founder of Berrykin Marketing
“It’s a really interesting topic. Whilst I use AI to help with research and admin tasks, I agree. Here, it can speed things up, but I wouldn’t be using it to make decisions from a business or marketing perspective; this is really frightening. From a marketing perspective, it needs to understand goals, KPIs, and the client, but it’s just scraping the internet. Yes, by inputting all the correct information you’d get an answer, but far too many are putting generic information in and therefore getting generic information in return.
“You need to be extremely targeted and tailored to use AI. It takes time to input this personalised information, so why not write it yourself and leave AI to specific admin tasks instead?
If your team removed AI tomorrow, would your core decisions actually get worse, or just slower?
“If our team removed AI, our work may be slower, but our core decisions wouldn’t be worse. Commercial judgement comes from experience, and AI just speeds it up.”
Jakub Lenski, Co-Founder of Saber
“This is something we strongly agree with. At Saber, our ethos is “Decisions stay human. Execution gets automated.” AI demonstrably speeds up processes, and our business has been built on the ability to process volumes of work in hours that would normally take weeks. What it can’t replace is human judgment and guidance. For AI to be effective, processes have to be clearly defined with carefully thought-out rules and guardrails, and data needs to be organised in an intentional way.
“It is these fundamentals that make in-house builds difficult to execute for teams new to AI. Human-in-the-loop has to remain a key part of any AI-assisted process, which can become cumbersome and sometimes even slower than the manual process if not implemented properly. In regulated industries like the ones we operate in, that is especially crucial.”
If your team removed AI tomorrow, would your core decisions actually get worse, or just slower?
“Our decisions wouldn’t necessarily get worse; they’d just get slower, and time would be taken by repetitive manual tasks related to decision-making, like resource gathering and finding, rather than being spent on tasks that deliver value for our clients.”
Carlos Esquivel, Managing Director (Technology) and Co-Founder of Journi
“AI rarely fixes weak judgement, and it can scale mistakes as efficiently as good decisions! But the problem starts with focussing on the technology before understanding the challenge.
“Many founders feel pressure to adopt AI because they fear falling behind. The useful question is not ‘What can AI do?’ but ‘What needs fixing?’ e.g. If customers wait two days for a quotation, identify what causes that delay. AI might help interpret an enquiry, but a clearer form or better use of existing systems could solve it more cheaply.
“AI can help people explore options and challenge assumptions, so its value goes beyond speed. But someone still needs to define success, check the output and own the decision. Automating a poorly understood process does not make it a better process. It makes its weaknesses harder to ignore.”
If your team removed AI tomorrow, would your core decisions actually get worse, or just slower?
“Mostly slower: AI helps us explore and test options, but our core decisions depend on expertise, evidence and accountable human judgement.”
Fiorela Imerai, SEO Account Director at Wildcat Digital
“The problem is that too many businesses are adopting AI without understanding what they’re actually trying to achieve. In SEO, for example, AI has made it incredibly easy to produce content, analyse competitors and identify opportunities at scale. But more content doesn’t automatically translate into better rankings, stronger customer engagement or increased revenue. I’ve seen how easy it is for businesses to confuse productivity with progress. You can automate an entire workflow, but if the underlying strategy is flawed, you’re simply executing the wrong approach faster.
“The businesses that will benefit most from AI are the ones with experienced people who understand their customers, challenge the outputs and know when human judgement matters. AI should make good strategists more effective, not replace the need for strategy altogether.
If your team removed AI tomorrow, would your core decisions actually get worse, or just slower?
“AI has become an important part of how we work, helping us improve efficiency and make more informed decisions. Without it, some processes would take longer, but our strategic decisions would still be driven by expertise, data and a clear understanding of our clients’ business objectives.”
Fernando Pedraz, CEO at Reveni
“If you cannot explain the problem you are solving, adding AI will not make your strategy clearer. It will help you produce more work around an unclear objective.
“At Reveni, when a merchant asks for something, we first question whether that request addresses the underlying problem and whether other merchants share it. That judgement has to come before choosing the technology.
“AI makes it easier to build, but building faster is not the same as building something useful. Founders still need to decide which problems deserve investment, what success looks like and whether the result works for customers. A polished output can make a weak idea look convincing. The responsibility for challenging it stays with the team.”
If your team removed AI tomorrow, would your core decisions actually get worse, or just slower?
“Without AI, execution might be slower, but our core decisions should still stand up because they are grounded in merchant problems. We also built our scaffolding and database before AI, which prevented us from vibe coding the most critical part of a business meant to scale.”
Pavlo Kharmanskyi, Founder of OnlyMonster
“I agree with the diagnosis, but I think it misses something about how AI is changing software development. I’ve been writing code since I was 14, and one thing I’ve learnt is that most technical problems are really problems of architecture, logic or understanding what you’re trying to build. AI hasn’t changed that.
“What it has changed is how quickly you can test your thinking. I can use an AI coding agent to build a working prototype, run tests, identify edge cases and explore several approaches in the time it previously took to develop one. That’s a pretty fundamental shift. The problem comes when founders confuse generating code with building reliable software. An LLM can produce something that looks perfectly functional while introducing security vulnerabilities, unnecessary dependencies or architectural problems that become expensive later.
“AI gives us much faster feedback loops. Whether we learn anything from them is still up to us.”
If your team removed AI tomorrow, would your core decisions actually get worse, or just slower?
“Both, actually. We’d move slower, but we’d also lose the ability to test as many ideas and technical assumptions before committing to a decision.”
