Interval

• —Most journaling apps are generic and don't target anxious attachment loops (rumination, reassurance-seeking, difficulty sitting with uncertainty).
• —Attachment-style mental health apps often feel clinical/cold, which creates friction for users who need warmth and safety.
• —Privacy is a prerequisite: users won't write honestly if they don't trust where their data goes.

I interviewed 12 people who self-identify as having anxious attachment.

• —Users wanted prompts that name the loop ("what are you seeking right now?") instead of generic mood check-ins.
• —Many described "checking behaviors" (messages / social feeds) as the main trigger pattern they want to interrupt.
• —The biggest adoption barrier was fear of judgment + fear of their writing being exposed.

A warm, offline-first journaling flow designed around anxious attachment patterns.

• —Structured prompts tailored to rumination / uncertainty / reassurance
• —Mood + behavior tagging that makes patterns visible over time
• —Gentle reminders (opt-in) that nudge reflection without shame
• —Importantly, wanted to add an AI chatbot to the app to help dig deeper into the user's thoughts and feelings -- but that's for later work when I am able to build an LLM that lets users not emotionally rely on it, which would thus put them in a position to actually deal with their thoughts and feelings, instead of using the app as a crutch.

• —Built with Expo + React Native to ship cross-platform quickly
• —Offline-first storage using local SQLite
• —Optional sync is intentionally deferred until security requirements are met

AI was used as leverage for speed and decision-making -- not for analyzing user journals. In addition to Figma AI for early wireframes, I used Google Stitch to rapidly explore higher-fidelity screen concepts across every surface of the app.

• —Figma AI generated a lo-fi walkthrough early -- validated the 4-tab structure fast before investing in high-fidelity screens.
• —Google Stitch wireframes explored the full app surface: journal prompts, archive logs, calendar views, reminders, and splash screen.
• —f5.ps generative typography exploration -- accelerated brand direction by testing identity energy quickly (even when the first mark failed).
• —AI was used for ideation + iteration, not for analyzing user journals.

Logo exploration (f5.ps) -- failed but informative

1/7

Interval is currently a concept MVP and is live on Replit for review/testing. I'm using it to validate the core flow (prompt → entry → tags → archive/calendar) and gather early feedback before any broader release.

• —Status: closed concept testing (not a public launch yet)
• —Security-first plan before shipping widely: privacy hardening, threat modeling, secure storage choices
• —Early signal: attachment-specific prompts are the standout differentiator vs generic journaling

still working out some kinks on this app!

• —Building solo meant wearing every hat: researcher, designer, engineer, PM.
• —Hardest tradeoff: resisting feature creep and keeping the scope tight around the core loop.
• —Offline-first added complexity but was non-negotiable for trust.