Epowar

AT A GLANCE

Personal safety, read from the wrist

• Sector — Personal safety and health tech
• Platforms — iOS app with a companion Apple Watch app
• What we did — Product design and development, delivered in 8+ weeks
• Sensor integration — Heart rate and motion read continuously from the user’s smartwatch
• Core features — Real-time journey tracking, automatic attack detection, alerts to chosen friends, secure evidence recording, audible alarm

THE CHALLENGE

A safety product that needs pressing often doesn’t get pressed

Conventional personal safety products share one point of failure: someone has to activate them. A rape alarm or a panic button only works if the person carrying it has a free hand, enough warning and enough presence of mind to use it, and in the moment an attack begins none of those can be relied on. Removing the button was the starting point for the whole product.

Whatever the user does still have to touch has to work in the worst conditions: one hand, poor light, a body in motion and a raised heart rate. Controls had to be few, large and reachable with a thumb, with nothing important hidden behind a menu or a sentence that needs reading.

Automatic detection solves the activation problem and creates a new one. A safety app that cries wolf gets muted or deleted, so detection built on heart rate and motion has to tell a sprint finish, a stumble or a jolt on the bus apart from genuine distress. That is why the models were trained on real data rather than on assumptions about how distress looks.

Then there is everything that has to keep running when nobody is looking at the screen: sensors read continuously, a location fix held, the app alive in the background, all on a battery the user still needs at the end of the night, with alerts leaving the phone over whatever signal is available. The app also has to be calm to browse. Someone downloading a safety app is already worried, so the design had to feel steady and matter-of-fact rather than trade on fear.

WHAT WE DESIGNED

An iOS and Apple Watch app built around the journey

The journey is the object everything else hangs off. A user starts one before setting out; from that point the watch supplies heart rate and motion, the phone acts as the hub, and chosen friends can follow the journey in real time. Tracking is reciprocal — you can follow their journeys as they follow yours — which makes it a shared habit rather than a feature you only touch when frightened.

The alert is a coordinated response rather than a single action. When the models detect signs of physical distress, the app alerts the user’s friends, records evidence securely and sounds an alarm, the alarm doing double duty as a deterrent and as a signal to anyone nearby.

Splitting the app across two devices meant deciding what belongs where. The watch is the sensor and the fastest surface to reach in a hurry, so it carries only what has to be immediate. Setting up trusted friends, reviewing past journeys and granting the health and location permissions the app depends on all live on the phone, where there is room to explain what is being read and why.

WHAT SHIPPED

An automatic response instead of a button

Epowar shipped as an iOS app with a companion Apple Watch app, delivered over 8+ weeks of development. The released product tracks journeys in real time, uses AI models trained on real data to detect signs of an attack from heart rate and motion, alerts the user’s chosen friends, records evidence securely and sounds an alarm.

No download, rating or funding figures have been published for this project, so we quote none. It is also worth saying plainly: no app can prevent an attack. Epowar was designed to shorten the time between something happening and someone else knowing about it.

FREQUENTLY ASKED

Commissioning a safety or wearable-connected app

How do you design an interface for emergency use?

You assume the worst conditions and work backwards from them: one hand, poor light, movement, and a user with no attention to spare for reading. In practice that means very few controls, large targets and no nested menus. Better still is to remove the interaction altogether — Epowar’s detection runs automatically, so the app does not depend on anyone reaching their phone.

How do you design apps that connect to wearables?

Start from what the sensors can actually tell you and how often, then design the experience around that rather than the other way round. For Epowar that meant heart rate and motion from an Apple Watch, read continuously through a journey, with the watch as the sensor and the phone as the hub for setup and history. Battery life, background running and the handover between the two devices are design constraints as much as engineering ones.

How do you build user trust in a safety product?

Accuracy first: training detection on real data matters because one false alarm costs more trust than several correct ones earn. Then honesty — being clear about what the app reads, what it records, where that goes and what it cannot do. A safety product should under-promise rather than imply a guarantee it cannot keep.

Related services

Safety products are judged on the moments that matter most, and those moments are pure interaction design. If your app has a critical flow that has to work first time, under stress, a free UX audit will tell you honestly whether it does. We work with founders on exactly this kind of problem as an app design agency in London — get started if you would like to talk it through.