Product Design · iOS · 2025

Resolv: AI Couples Coach
Couples therapy costs $200 a session and takes weeks to book. I built the in-between.

Role

Sole Designer · Product, Research, Systems, Brand

Timeline

Launched in iOS app store December 2025

Platform

iOS · React Native

Resolv: intake counselor, joint session, and resolution screens

The Problem

Couples in conflict don’t have good tools

$200 per session

The average session costs $150–$250. The math ends the conversation before it starts.

2–6 weeks to first appointment

By the time they sit with a therapist, the original fight is already a scar.

Conflict doesn’t wait for an appointment

When couples are mid-argument, they need guidance in the moment — not two weeks later.

Why I built this

Couples therapy changed my relationship. After 13 years with my partner, I’d seen what a structured, guided conversation could unlock — and how rarely couples actually have access to it. When I looked at the AI therapy space, almost everything was built for one person. The two-person problem — competing realities, no neutral ground, conflict that doesn’t wait for an appointment — was almost entirely ignored.

That gap felt worth building against. And honestly, I also wanted to learn to build. This was a real problem, not a toy, and that made all the difference.

I would literally feel getting lost in a conversation, anchoring myself to phrases that in the big scheme of things were noise.

Early user describing their pain point when in conflict with their partner

What I Designed

The benchmark wasn’t other AI products. It was a good couples therapist.

What didn’t work first

My first version worked differently. Each partner submitted their side of the story separately, and the AI analyzed both responses and figured out next steps. It was efficient. It also felt like a verdict — one person's account weighed against the other's, with an algorithm deciding who was right.

When I tested it on my own relationship, the problem was obvious. So I started studying how actual couples therapists work. They don't collect forms and render judgment. They sit with both people in the room, slow things down, and create the conditions for each person to feel heard before they can hear each other. That structure became the model. A private intake for each partner first, then a joint session guided by what each person shared. Not a chatbot asking questions. A session with a shape.

Private before joint

Real therapists don’t put two people in conflict in a room cold. Each partner speaks privately first — so they feel heard before they have to hear each other.

Resolv private intake: one partner in a one-on-one session with the AI counselor

Structure creates safety

A good therapist knows what kind of session this is before it starts. Repair, validation, understanding, resolution — each follows a distinct therapeutic arc, invisible to the couple but guiding every exchange.

Joint session UI: both partners in the room with the AI guiding the conversation

End with something tangible

Sessions that end in air don’t stick. The resolution card captures what both partners agreed to, in their own words, with a date attached.

Resolution card showing Partner A and Partner B commitments with dates

Built and shipped

After 1.5 months of vibe coding, bug fixing, and navigating App Store guidelines, Resolv is live in the App Store

View on the App Store
Resolv on the App Store: app listing with ratings and description

Outcomes

200+

Downloads since launch

5.0

Average rating on the App Store

User quote

“By using Resolv, I was able to slow my reactions down, recognize the pain that my partner was in and offer a way that I could support them.” — Resolv user

Live

Launched December 2025 · iOS App Store

What I learned

Building this changed how I design

Designing for two is a different problem entirely.

Every UX pattern I’d internalized assumes one person with one goal. Two people in conflict have competing goals, competing versions of the same event, and different needs in the exact same moment. The private intake wasn’t an extra step. It was the only way to make the joint session feel safe enough to be useful.

Ship it, then design it.

Vibe coding collapsed the gap between mock and reality in a way that changed how I prioritize. The resolution card only became obvious once I watched a real session end without one. You can’t design the ending until you’ve lived through the middle.

The benchmark matters more than the feature list.

The question I kept coming back to was: what would a good couples therapist do here? Not a good chatbot. That single reframe filtered almost every product decision — session structure, AI tone, what to surface at the end. Picking the right benchmark is itself a design decision.

Next case study
Amazon Customer Service Hub & Funnel Systems Design · Research · Scale