Innovation Through Alignment

Why technology rarely fails — but misalignment often does

In dementia care, innovation is often associated with new technology: smarter sensors, better AI, more data. Our experience in daily practice tells a different story.

Innovation rarely fails because technology is insufficient.It fails because systems are not aligned with real life.

Working closely with care organisations, professionals and technology partners, we have learned that even advanced solutions can create unrest instead of calm when they are not attuned to daily routines, human behaviour and care workflows.

That insight lies at the heart of our contribution to the World JAIN Challenge 2026 — and it explains why alignment, not technology, is the real driver of Human Friendly AI.

The hidden impact of “out‑of‑the‑box”

In complex care environments, technology almost never works “out of the box”. Not because it is poorly designed, but because care is not a controlled environment.

• Every location is different.

• Every resident behaves differently.

• Every night unfolds in its own way.

One of the most underestimated risks is false alerts. False alerts disturb residents’ sleep, increase stress and cognitive load for care teams, and gradually undermine trust in the system. In practice, they often cause more harm than having no alert at all.

This led us to shift our focus. Instead of asking how to detect more, we started asking a different question:

Where does misalignment occur — and what does it do to people?

Innovation through alignment

What we see in practice is that misalignment happens on several levels at once:

• between people and technology,

• between physical reality and system design,

• between assumed and actual behaviour,

• and between systems that need to work together.

Care professionals should not have to think about configuration during a night shift. Sensors should not depend on perfect placement. Behaviour should be observed, not assumed. And in a multi‑vendor reality, interoperability is not optional — it is essential.

True innovation begins when these forms of misalignment are addressed together.

A small insight with a big effect

One of our most instructive lessons came from a simple observation.

By analysing movement patterns and listening to care professionals, we noticed that residents often move very close to their bed when getting up at night — for example when going to the bathroom. Traditional detection models assumed more distance.

By shortening the bed detection zone, detection became faster, more accurate and far less prone to false alerts.

No new hardware.

No new AI model.

Just better alignment with reality.

This is what Human Friendly AI looks like in practice: small, grounded insights that create disproportionate impact in calm, safety and trust.

Privacy by design, not as an afterthought

In dementia care, technology must be present without being intrusive. That is why our approach avoids cameras and wearables, using imaging radar instead — detecting movement patterns while respecting privacy by design.

Human Friendly AI is not only about what technology can do, but also about what it deliberately does not do.

Why we share this

We see our finalist position in the World JAIN Challenge 2026 not as a podium, but as a responsibility. The challenges in care are too great to keep practical knowledge to ourselves. By sharing insights from daily practice, we hope to contribute to a broader understanding of how technology can genuinely support calm, dignity and safety.

Not as a blueprint to copy, but as a way of thinking that keeps people — not systems — at the centre.

The real winners of good care technology often do not even know that the technology is there.

We consider that the greatest compliment.

More about JAIN