Steve Jobs said something years ago that explains very well what is happening with AI right now: “You’ve gotta start with the customer experience and work backwards to the technology. You can’t start with the technology and try to figure out where you’re going to try to sell it.”
This is exactly the mistake many companies are making with AI right now. They start with AI. We need AI. Let’s figure out where we can integrate it. We need an AI agent, an AI chatbot, an AI assistant, AI in our product, AI in our internal processes. Does it actually make anything better? Does it solve any real problem? Does the customer even need it? Quite often no one seems to care. There is amazing AI technology, so the thinking becomes: let’s find somewhere to put it.
This is completely backwards. Companies are starting with the technology and then trying to invent a use case around it without properly understanding the problem, the customer or even whether AI improves anything at all.
You should first find the actual customer problem. What is broken, what is expensive, what takes too much time, what do people hate doing, what is still being done manually, and what does the customer actually want to achieve? Then you figure out how to solve it. Maybe AI is the best technology for it, maybe it is not. The goal should be to solve the problem, not to find a reason to use AI.
This is also why we have so many AI products right now that look amazing in demos but people stop using them after a few weeks. A cool AI demo is not a business. Many of these products are basically gadgets. They are cool to try, play with and show to other people, but when the excitement wears off, there is often very little real-life practical value. The important question is not whether AI can do something impressive. The question is whether it solves a real problem well enough that people actually want to keep using it.
Generating a presentation is cool, summarizing a meeting is cool, talking with your documents is cool, building an AI agent that can click around websites is cool. But what real problem does it solve? How much money does it save? How much time does it save? Does it actually make the complete process better? Would somebody still pay for it after the AI hype disappears?
Customers normally do not care about AI itself. A company does not really want AI cybersecurity. They want fewer security problems and they want vulnerabilities fixed. They do not want an AI vulnerability management platform because AI sounds cool. They want somebody to tell them which 20 vulnerabilities actually matter instead of giving them another Excel file with 50,000 findings. A salesperson does not want AI because it is AI. They want to spend less time filling CRM fields and more time selling. A support team does not care which LLM is behind the chatbot. They want fewer repetitive tickets and customers getting correct answers faster.
The result matters, not the technology.
I think this is one of the biggest problems with the current AI market. A lot of companies are technology-first instead of customer-first. They have already decided that AI must be the solution before they even properly understand the problem. The discussion starts with “Where can we integrate AI?” instead of “What problem are we trying to solve?” That small difference changes everything.
If you first decide that AI must be used, then you are already limiting yourself. You are no longer looking for the best solution. You are looking for a justification for using the technology you already picked. This is exactly what Steve Jobs was warning about.
For the last few years, adding “AI” to almost anything also made it easier to get attention, funding and customers, so naturally everyone started adding AI everywhere. But eventually somebody has to ask whether the thing is actually useful. Not whether the technology is impressive, but whether the customer experience is better because of it.
Another problem is that many AI companies are building entire businesses around something that is actually just a feature. Document summarization is a good example. A few years ago you could probably build a whole startup around it. Now Microsoft does it, Google does it, OpenAI does it, Anthropic does it and almost every productivity platform is adding it. So what is your business when everyone has the same feature? “We use AI” is not a competitive advantage. Everyone has AI.
The real advantage is understanding a specific customer, a specific industry and a specific problem better than everyone else. AI should make the solution better, but AI itself should not be the whole reason why the product exists.
There is another thing that AI demos often hide. Real companies are messy. They have old systems, badly designed processes, security requirements, compliance, contracts, approvals, different departments, outsourced services and ten different tools that do not properly talk to each other. Then somebody comes with an AI tool that saves one employee ten minutes, but now the company also has another subscription, another login, another vendor, another security review, another integration, another contract and another system somebody has to maintain. Did we actually improve anything?
This is why you have to look at the complete customer experience and not one AI feature. Maybe the real problem is not that somebody needs AI to read 10,000 alerts. Maybe the real problem is why they are receiving 10,000 useless alerts in the first place. Maybe the solution is better prioritization, better asset data, better processes and better ownership. Then AI can make that solution much better.
That is how I think AI should be used. First solve the real problem, then use AI where AI actually makes the solution better. Not AI first and problem second.
I also think the best AI products in the future will stop talking about AI all the time. Nobody buys Netflix because of the video codec. Nobody uses Google Maps because they are excited about GPS technology. Nobody really cares what technology is working behind the scenes if the product solves their problem. AI will become the same.
The best AI product might be one where the customer barely thinks about AI at all. They just notice that something that previously took four hours now takes five minutes, something that previously required six systems now happens in one place, or something that previously required manually checking 50,000 findings now gives them the 20 things they actually need to fix.
That is value.
Steve Jobs understood this a long time ago. Start with the customer, understand what they actually need, think about what incredible benefit you can give them, and then work backwards and choose the technology. A lot of AI companies are doing exactly the opposite right now, and I think that is one of the main reasons why so many of them will fail.