Tell Me a Story
Snapshot
After graduating I was contacted to continue a neonatal project I'd worked on as a student, taking it from a paper prototype to a working voice-recording prototype, tested with six parents who had lived through NICU care.
Tell Me A Story lets parents record stories and messages for their baby in a Neonatal Intensive Care Unit, so their voice reaches the cot when they can't be there.
Problem
In a NICU the bond between parent and newborn is disrupted exactly when it matters most: you go from birth, to touching your baby through an incubator, to leaving them overnight. Parents get updates but have no way to reach their baby, and a parent's voice is one of the few things known to support both brain development and bonding.
I was brought back in to continue the work, which meant starting from what already existed rather than a blank page: a DJCAD and School of Medicine study on the Ninewells neonatal ward, and a 2023 third-year project I'd built with Pavlin Petev: the paired work that first proposed recording stories for playback at the cot. Seven prototypes from that phase had been shown to six experts, pointing at four areas to develop: recording support, memories and metadata, a custom avatar, and signposting to support information.
Constraints
Playback could never be guaranteed. As the senior neonatal nurse put it, “there would be intent to play the audio, but no guarantee, clinical duties come first.” The design had to set that expectation rather than hide it.
Safeguarding had to be visible.“We'd need a way to check it's for the right baby”, plus open questions on NHS system integration and information governance.
The prototype had to actually record audio, which pushed me onto Bubble.io. It handled the recording but restricted the design everywhere else, and being web-based added a lag between screens.
Worth noting this was pre-AI. There was no assistant to lean on for the parts Bubble.io made difficult, so getting functional voice recording, pause, restart and playback working took far longer than the design itself, and every workaround had to be found the slow way.
Continue the paired concept, don’t restart it
The 2023 project already had a defensible core, record a story, send it to the ward, play it at the cot. Rather than redesign that, I treated it as the starting position and spent the time on what it still lacked: a real recording flow, an emotional register, and evidence from parents.
First move was a paper prototype, to make the concept tangible enough to react to before committing to a build.
Split recording into a story and a quick message
Testing the paper version made it clear that one recording flow couldn't serve both intents. A parent reading a book needs to pause, restart and hear it back. A parent saying goodnight before they leave needs one tap.
So the flow forks: Special Story carries the full controls, Sweet Message behaves like a voice note. Which one you pick depends less on the feature than on how much you have left in you that day.
Design it away from anything medical
The Ninewells NICU is deliberately softened with illustrations by Freya Cumming, and I took the same line: nothing that reads as clinical or cold. Red was ruled out entirely, in that environment it means emergency.
The reference point became children's books, and the design settled on a night sky with a glowing yellow. The onboarding does the expectation-setting the nurse asked for, and one of her phrases, “anything that leaves that little heartstring of attachment”, made it into the app itself.
Recruit parents who’ve been through it, never those still in it
Asking a parent to evaluate a prototype while their baby is in intensive care would have been invasive, so current NICU parents were ruled out. Recruitment was limited to parents within five years of their experience, recent enough to recall clearly, and to the UK, since the NHS was the target context.
17 people showed interest, 11 completed ethics, 6 tested : a 35% conversion with no incentive offered beyond interest in the project. Sessions ran up to an hour as semi-structured interviews around the prototype. All six participants were female and UK-based.
Shipped
A working high-fidelity prototype in Bubble.io with functional voice recording: onboarding, the story-type fork, recording with playback, a Stories page (Daytime, Bedtime, A Story I Love), and a Home page of storytelling tips.
Alongside it, a full research report, protocol, participants, findings, emerging themes and recommendations, plus the nurse-side requirements that fell out of scope.
Outcome
Feedback was overwhelmingly positive. Parents said they would have valued it during their own stay, “just knowing they could still hear your voice would have helped”, and the tone landed as intended, described as “communicated in a compassionate and sensitive way”, in contrast to existing neonatal apps that “wasn't a very visually pleasing system”.
Testing produced six themes with design implications, bonding through audio, the need for normalcy, language, emotional load, guilt, and visual design, plus five recommendations and a feature shortlist (a priority “special request”, and stories from wider family). Three further participants were tested after the internship ended.
Reflection
The most useful finding was the one I hadn't designed for: parents didn't always know what to say. “I felt really self-conscious about talking or reading,” one said; another wanted “something to jog your brain a little bit so you get comfy.” The app made recording possible without making it easy to start, prompts would be the first thing I'd add.
The nurse-side interface also fell out of scope as the parent flow grew, and it's the half that decides whether any of this works on a ward. My part is complete; the wider initiative has slowed since, so I treat it as paused rather than abandoned.