The empathy deficit: the mistake most founders make with AI

Every founder knows the pressure: ship fast, stay lean, beat the competition to market. AI has supercharged that instinct with cheaper tools and shorter development cycles. The temptation to automate everything is overwhelming, but there’s a problem hiding in that speed. When executives were asked which skills will be essential in the next two to three years to compete in an AI-driven world, 52% championed critical thinking, while only 17% rated empathy as essential. That ratio should worry anyone building a product.

To be clear: when we talk about empathy in AI strategy, we don’t mean being nice. We mean cognitive empathy, the strategic ability to model how your users actually think, decide and behave. It’s what helps you spot patterns in genuine human behaviour that your data hasn’t captured yet. Without it, you’re not just building blind; you’re building something expensive that nobody wants.

We call this ‘The Empathy Deficit,’ and for startups, where every pound of runway matters, it might be the most dangerous blind spot in your AI strategy. The assumption that soft skills don’t move the needle on ROI is exactly the kind of thinking that leads to products nobody uses. Treating empathy as optional in the age of AI is a catastrophic business error.

Feeling sceptical? Look at the economics.

Why ‘soft skills’ are a hard advantage

When you’re building an AI product, empathy isn’t a luxury, it’s about making better decisions when you don’t have all the answers. It’s the capability that helps teams recognise unspoken frustrations, unmet needs and behaviours that haven’t yet surfaced, the things that determine whether your product gets adopted or abandoned. Logic alone can’t see those things.

And here’s what makes this urgent: as LLMs become more sophisticated, critical thinking, processing information to reach a logical conclusion, is becoming a commodity. It’s cheap, instant and abundant. When any resource floods the market, its value drops.

The value of deep human understanding, meanwhile, goes up because it’s the one thing machines cannot simulate effectively. For founders, this reframes the entire competitive landscape. Your ‘Human Moat’ is no longer how to solve the problem (logic), but knowing which problems are actually worth solving (empathy). That’s where EQ (emotional quotient) becomes the differentiator between a product that scales and one that stalls.

What the deficit actually costs

There’s a pattern we see repeatedly: businesses are excited about AI, but when they see the price tag, the first things cut are the ‘soft stuff’, like experience design and change management. It’s a natural instinct, especially when cash is tight. But it’s a costly one. If your product isn’t grounded in how people actually behave, you’re gambling with every pound you’ve raised.

And the consequences compound quickly. The Empathy Deficit doesn’t just produce products that underperform; it erodes customer experience and, left unchecked, ignites the kind of brand crises that no amount of technical excellence can undo.

When automation gets it wrong

Prioritise speed over understanding and you don’t just build something that doesn’t work, you build something that actively pushes customers away. McDonald’s partnered with IBM to automate drive-throughs with voice AI, aiming for speed and efficiency. But the design didn’t account for the chaotic reality of ordering food for a family, with background noise, accents, even kids shouting in the back seat. Customers ended up trapped in loops, arguing with a bot over basic requests. The partnership was eventually scrapped. The Empathy Deficit cost: millions in wasted R&D, lost efficiency gains, and a very public reminder that speed means nothing if the experience falls apart.

It’s not an isolated case. A DPD customer realised the company’s AI support bot couldn’t handle his missing parcel issue with any nuance, so he tested its limits by convincing the bot to swear at him and write a poem about how terrible DPD was. The bot obliged. The fallout went viral. The Empathy Deficit cost: forced shutdown of automated support channels, global media coverage, and lasting reputational damage.

Empathy as a guardrail against risk

One biased algorithm or insensitive interaction can undo years of brand-building. And for a startup, where brand trust is still fragile, the stakes are even higher. Empathy is the quality control layer that prevents AI from scaling bias.

See Also

Facial recognition is at a crossroads: trust and privacy are key to its progress

Consider when Microsoft announced a feature that took constant screenshots of users’ laptops to create a “photographic memory.” They viewed it as a productivity hack; users viewed it as spyware. The backlash forced a delay and complete re-engineering of a flagship feature because the design failed to anticipate the very human fear of surveillance. The Empathy Deficit cost: delayed launch, wasted development investment, and a very public erosion of user trust. If Microsoft can get this wrong with their resources, imagine the damage for a startup that can’t afford a second chance.

How to build it in without slowing down

The good news is that the process for embedding empathy already exists – it’s design thinking. It’s nothing new, but in the rush to ship AI products fast, it’s been quietly sidelined. You don’t need to invent a new methodology; you need to stop skipping the one that works. Start with the delivery phases you already have:

• Before you build: surface hidden user needs first. Talk to people. Observe how they actually behave. Then translate what you learn into user stories your team can build against. If you don’t have someone trained in this, like a UX researcher or experience strategist, bring one in. This isn’t work you can hand to a generalist and hope for the best
• While you build: use AI to accelerate human-centred design, not replace it. Teams can explore assumptions, pressure-test ideas and prototype interactions before writing production code. This surfaces friction points early and focuses testing where it matters most
• After you launch: build in continuous feedback loops. Post-launch sentiment analysis and rapid iteration ensure your product evolves with human behaviour, not against it

The inevitable question: ‘Won’t this slow us down?’ It won’t. It accelerates delivery by preventing the expensive mistakes that happen when you build the wrong thing fast. Teams that spend two to three days validating assumptions upfront typically save three to four months of rework downstream. Speed without direction isn’t velocity; it’s accelerated waste.

Empathy isn’t fluff – it’s what stops your AI strategy from becoming your most expensive mistake. Founders who recognise this early gain a strategic edge, while their competitors are still building fast in the wrong direction.