Designing and building Sift, an iPhone notes app I can ask about anything I've captured, from quick text notes to voice memos and links
Challenge
The Notes app on my iPhone had become a dumping ground for ideas, thoughts and recommendations captured in the moment, as text, voice memos and links. Most of them are spur-of-the-moment ideas I want to note down before I forget them, captured quickly while I'm watching a video or doing something else. Written that quickly, the notes rarely carry the context that makes them worth revisiting. The question I then asked myself was: what if I could chat with the notes I had captured in the past and get refined context and ideas back, drawn only from those notes? I also wanted to push my use of AI beyond my comfort level. So the goal was to build Sift with AI as a real app I use myself, not a solution that only works on paper.
Approach
The idea was not to reinvent the wheel when it comes to capturing notes on Sift. The Notes app on my iPhone already does that well, so Sift builds on what already works and fills the gap. I explored how other tools solve similar problems and used what I found to shape a solution that fits mine. I also chose to get to an MVP quickly rather than build something perfect from day one. To build Sift, I used Claude Code, working in plain language. I followed a process I've learned over time as a designer building with AI: get the idea down, have at least a version of the designs ready, and give the AI as much context as possible. Then ask it for a plan, including the tech stack, and keep building and iterating until the app takes shape.
Solution
The app is an MVP of a notes app powered by AI. The capture side is a stripped-back version of the Notes app on my iPhone. It keeps only the features I use: text with formatting, photos and videos, links, and voice recordings that get transcribed. Everything else the Notes app offers, like math results and sketches and markup, is left out because I never use it. The app then adds a chat on top of capture, with two modes: Ask this note, for a single note, and Ask notes, for everything I've captured. Notes are processed in the background, so I can ask for what I need in plain language, like a list of every movie I've saved. Answers come only from my notes and point back to the original note. If the answer isn't in my notes, the app says so.
Outcome
The result is a working MVP on my phone, which I'm refining to fix bugs and improve the look and feel and the answers it gives. Being able to ask my notes has given me the confidence to write things down, knowing the context will never be lost. Building with AI has helped me prompt better and understand how products are developed. As a product designer, I already design within technical constraints, but doing the development on this app myself brought me closer to those constraints, rather than being told what they are by a developer I'm handing design work over to. I've built the idea, and I'll take what I learned into the next thing I build. Next, I plan to add new features and potentially grow the app into a more capable AI app.