inspiration
pip tales came from something very personal. my little sister loves stories. she loves telling them, changing them midway, asking wild questions, and hearing the same characters come back in new adventures. but real life is busy now. people are tired, schedules are packed, and the slow magical act of sitting down and building a story together happens less and less.
i did not want her free time to get eaten up by the usual brain rot loop of endless short videos and shallow content. i wanted something that felt alive, creative, and comforting. something where her favorite characters could actually show up, notice her world, talk back, and turn her own imagination into something lasting. that became pip tales.
after building the first demo, i tested it with my sister in a real session, and that was the moment it clicked. she wasn’t just using an app. she was directing her own little world.
what it does
pip tales is a live storytelling companion for kids.
a child opens live mode, points the camera at the world around them, and starts talking. their favorite character responds in real time, reacts to what it sees, and helps the child turn ordinary moments into a playful adventure. the app saves the important beats from that interaction, then transforms them into a scrapbook-style story with cartoon visuals and short narrative scenes.
the result is not passive content. it is a story the child helped create.
how we built it
we built pip tales as a local-first next.js app with gemini at the center of the experience.
the live mode uses gemini live for real-time multimodal interaction, letting the character respond to voice and camera context during the session. after the conversation, gemini extracts the key moments, writes a lightweight story arc, and generates scene prompts for each beat. those prompts, along with captured session images, are passed into nano banana to turn the child’s real world into a stylized cartoon adventure. everything is then assembled into a scrapbook experience stored locally on-device for an mvp-friendly flow.
we focused on keeping the loop simple: live conversation, key moment capture, story generation, visual transformation, scrapbook playback.
challenges we ran into
one challenge was making real-time interaction feel smooth enough to be magical rather than awkward. handling live audio, camera frames, and session state together took a lot of iteration before the experience felt stable.
another challenge was consistency. when a child creates a character, that character has to feel recognizable across multiple generated scenes, even when the setting changes completely. keeping the visual identity coherent while still letting the story stay imaginative was harder than expected.
we also had to design around attention span. kids do not tolerate slow or confusing flows. if anything feels too technical, too delayed, or too repetitive, they leave instantly. that forced us to simplify the product aggressively.
accomplishments that we're proud of
the biggest one is that pip tales worked in a real demo with my sister. she immediately understood it, started improvising her own story, and kept pulling the character into new directions without needing any explanation. that was the product test that mattered most.
we are also proud that we showed pip tales to 20 kids and every single one kept coming back to it, building stories, replaying scenes, and asking to make “just one more adventure.” that level of repeat engagement told us this was not just a neat demo. it had real pull.
we are proud that pip tales turns ai into a creative companion instead of a content slot machine. it gives kids something to make, not just something to consume.
what we learned
we learned that children want agency more than polish. they care less about perfect graphics and more about whether the character feels real, responsive, and emotionally present.
we also learned that storytelling is a stronger interface than prompting. kids do not want to “use ai.” they want to talk, imagine, laugh, and keep the story going. once we designed around that truth, the product became much clearer.
we also saw that parents respond strongly to tools that pull kids away from low-quality passive media and toward creation, language, and imagination.
what's next for pip tales
the next step for pip tales is making the experience even more live and relational.
we want to introduce live teachers and richer video-native models so characters, guides, and learning companions can interact in real time without waiting for a scrapbook generation step afterward. instead of “talk now, generate later,” the story world itself should evolve live in front of the child.
we also want pip tales to expand from storytelling into guided learning adventures, where a child can explore reading, speaking, drawing, and emotional expression through characters they already trust.
Log in or sign up for Devpost to join the conversation.