Mindscape AI is a mental wellness app built around the habit of journaling. The user writes, the AI analyzes, and it learns enough to ask a better question tomorrow. I designed the conceptual iOS MVP end-to-end, onboarding, journaling, insights, and provider connection, over three months with the founding team.
The founders came to me with the core features in mind: AI prompts driven by sentiment analysis, mood trends, recurring theme detection, provider integration, a CBT-trained chatbot, and scheduling. The system had a spec, but what it didn't have yet was a design that felt earned. My job was to treat the feature list as input, not output.
The bigger tension wasn't between features, but between two users with opposite needs.
Someone using Mindscape on their own wants it to feel personalized. Someone whose therapist prescribed it wants structure they can show up to. The solo user would close it if the app felt too clinical. The prescribed user would close it if it felt too much like a wellness toy catered to aesthetics. The interface had to hold both at once without collapsing into either, while still feeling like something someone reaches for at 11 PM on a random Tuesday.
Four design decisions defined how the product feels.
Onboarding is where users start to grant trust, especially in a product that works with their mental wellness.
I split it into three different sections around identity, intent, and clinical context, so that each input maps to a piece of the experience that changes because of it.
AI is useful for analyzing patterns, surfacing themes, and suggesting prompts. But in the context of Mindscape, it's not useful as something the user confronts on every screen. An AI that's always present performs more intimacy than it has earned, and that's where trust breaks. So the prompt arrives at the top of the entry, tagged "based on your journals," and then disappears until the user searches for insights.
An earlier version gated insights behind linking a therapist. I flipped it, because Not everyone has a provider, and gating insights behind one would push away the user who just needs a low-friction place to write. It would also make the clinical side feel like the real product and the solo side a stripped-down preview.
The journal and insights are the user's by default. Provider connection adds to the experience instead of unlocking it.
Most journaling apps go bright and pastel. I had a vision for users opening Mindscape during nights, when thoughts swarm or something needs to be written down before it slips. The interface had to feel like a room with the lights low.
The full MVP flow was prototyped in Figma: onboarding through home, journal entry, viewing insights, provider connection, and profile.
The brief initially included a CBT-trained chatbot. However, this surfaced as a concern because framing the chatbot as therapeutic has to be held to a clinical standard, and one trained on CBT principles without licensed oversight is closer to a wellness tool wearing a lab coat. It would set an expectation the product cannot meet.
I considered AI-generated affirmations as daily push notifications, but decided to keep them in the app instead. A push that says "you are enough" hits differently when it interrupts a meeting than when it appears after the user finishes a journal prompt after a stressful day. Context has to be considered.
An earlier version gated the insights page behind linking a therapist, with prescribed journaling assignments built on top. For this MVP I cut the prescribed-assignment layer and kept the core journaling flow. Insights are the user's by default; provider connection adds to the experience instead of unlocking it.
The current connect-support screen lists providers. The next version handles intake and scheduling: the user filters by what they wrote about, sees availability, and books without leaving the app. The journal already knows what they've been working on, so the scheduler retrieves that context instead of asking again.
The brief also called for activity and sleep tracking. Pulled into the insights page, that data would let the app show what no digital journal can show on its own: that the bad week was also the week the user slept four hours a night 
The journal records what the user noticed. Wearable data physically records what they didn't.
I'd love to chat more about my process: