Designing the AI-Native
Infrastructure for Family Experiences
Why Now
The Market Moment
Family experiences are a marketplace category the internet tried and never fully solved. The reason isn't lack of demand. It's unit economics. Supply is fragmented and hard to aggregate. Matching is high-effort and low-frequency.
AI changes the unit economics that made it fail. That's the reason this moment is different.
Here's what broken looks like on the ground--->
My Role & The Constraint
Timeline:
January 2026 to present
Team:
3 engineers, 3 marketing members,
1 ux researcher, pm + uiux designer (me)
What I owned:
-
End-to-end product strategy and design
-
Motion design & implement (Cursor)
-
12 user interviews and 5 merchant interviews
-
Product concept video — Google AI Studio + After Effects
-
Partial frontend — UI, bug fixes, and interaction (Cursor + Xcode)
In the AI era, the product's job is to deliver an outcome — not expose a process.
Users shouldn't have to navigate to an answer. The answer should arrive.
This became the guiding principle for every major design decision on Youmigo — applied consistently on both sides of the marketplace.
The Design Lens:
Outcome-Oriented Design
For Families
Find activities, fast
Crawler Agent: Aggregates open web data for Day1 supply.
Discovery & Advisor Agent: Suggests perfect match & Delivers growth insights
Enjoy the experience, stress-free
Planning Agent: Auto-builds tailored weekly schedules.
Create events, simply
The traditional listing flow asks hosts to fill out fields one by one.
We replaced it with a Creator Agent that takes whatever input the host has — a photo, a voice note, an existing Instagram post, a flyer — and produces a bookable event.
The pivot
Early prototype testing with onboarded merchants revealed that roughly half found manual input too slow. They didn't want to start from scratch.
They wanted to migrate what already existed. This led directly to three additional input paths:
Import from Instagram Post
Import from Facebook Post
From Flyer — extract structured data from a host's existing promotional material
Manage events, effortlessly
Receptionist Agent: Handles inquiries & payments 24/7.
Advisor Agent: Predicts demand & optimizes pricing to maximize revenue.
Designing the Path to Supply
How to balance speed, cold start, and users' behavior change.
For Merchants
More Design Details
Light & Dark mode
Login
Onboarding
Filter
Outcomes
45 days after MVP launch
Reflections
The Workflow Shift
The Double Diamond still applies. But the speed of each loop has compressed dramatically. AI tooling means I can go from a rough idea to a testable, real-feeling prototype in hours. The bottleneck is no longer implementation. It's thinking clearly about what to build.
The motion I implemented in Cursor after finishing drafting in Figma. It has already been shipped. ---->
On thinking ahead — and knowing when not to
It's easy to get excited about where things are going. A2A infrastructure, AI agents booking experiences, the entire discovery layer moving off traditional search. I believe that's the direction. We're building toward it.
But across 50+ parent interviews, fewer than 5% had used any AI tool to find local activities. The behavior shift hasn't happened yet. Think ahead, build the infrastructure that will matter — but always come back to what users are actually doing right now. The vision justifies the architecture. The user research justifies the product.