CASE STUDY / META

CASE STUDY / META

Horizon Create

Horizon Create

Horizon Create

Building a generative AI game-creation app from the ground up

Building a generative AI game-creation app from the ground up

Building a generative AI game-creation app from the ground up

WHAT

A zero-to-one generative AI game-creation mobile app built using a conversational UI. It was being built alongside a companion web app, Horizon Studio.

MY ROLE

MY ROLE

Content designer across conversation strategy, creation and sharing product flows, and cross-platform terminology

CORE PROBLEM

CORE PROBLEM

Teams were working in silos, and foundational product decisions remained unresolved despite an initial eight-week development timeline.

KEY PROJECT CONTRIBUTIONS

KEY PROJECT CONTRIBUTIONS

  • Drove the conversation strategy through leadership approval and into model implementation.

  • Redesigned creation, sharing, and publishing flows to resolve ambiguity and unblock implementation.

  • Drove cross-platform consistency by establishing shared strategy and terminology governance.

Two Horizon Create mobile screens showing a natural-language game creation chat and a finished game

01 / CONVERSATION STRATEGY

01 / CONVERSATION STRATEGY

Creating a unified conversation strategy

Creating a unified conversation strategy

Creating a unified conversation strategy

THE INITIAL PROBLEM

Multiple disciplines were separately working on pieces of a conversation strategy for the AI model, but they weren’t collaborating, resulting in duplicate work and a disjointed user experience.

Multiple disciplines were separately working on pieces of a conversation strategy for the AI model, but they weren’t collaborating, resulting in duplicate work and a disjointed user experience.

Multiple disciplines were separately working on pieces of a conversation strategy for the AI model, but they weren’t collaborating, resulting in duplicate work and a disjointed user experience.

MY APPROACH

  • Identified and consolidated disparate conversation strategy efforts by meeting with cross-functional stakeholders and reviewing existing documentation and frameworks to understand the current landscape and technical constraints.

  • Established a unified strategic direction. I evaluated competing approaches, reconciled inconsistencies, and synthesized the strongest elements into a cohesive conversation strategy.

Illustrated process showing research, CXD and PMM guidance, product flows, and engineering constraints synthesized into a unified conversation strategy
Illustration of how a 5–10 minute generative-AI delay informed the principle of displaying visual progress inline in the conversation

IMPACT

  • Leadership approved the strategy. I led multiple rounds of review and aligned design and product leadership on the strategic direction.

  • Conversation strategy was incorporated into the model implementation. I brought the conversation designer and engineering team together to iron out practical constraints, fast-tracking internal user testing.

  • The web team adopted much of the framework and principles, adapting them to match their users’ specific needs.

02 / PUBLISHING FLOW

02 / PUBLISHING FLOW

Shaping the publishing experience

Shaping the publishing experience

Shaping the publishing experience

With the conversation foundations laid, I shifted focus to the actual in-product flows and interactions.

With the conversation foundations laid, I shifted focus to the actual in-product flows and interactions.

With the conversation foundations laid, I shifted focus to the actual in-product flows and interactions.

THE PROBLEM

The team lacked a shared understanding of what sharing and publishing meant, creating confusion in both the product experience and implementation.

MY APPROACH

  • Challenged the underlying assumptions. PMs proposed a long disclaimer to explain who could access a game because, technically, a privately shared game “draft” was considered “published” on the backend. I pushed back on exposing this backend terminology in the UI and partnered with my product designer to rethink the experience.

  • Advocated for distinct sharing and publishing actions. The lead CD on the app initially favored using “Share” for everything for simplicity’s sake. I argued that for clarity we needed two actions that reflected different user intentions:

    • Share: Send a game directly to someone.

    • Publish: Finish creating a game and choose its visibility.

    • Through iterative design reviews, my PD and I demonstrated how distinguishing these actions created a clearer experience, ultimately gaining alignment on the two-term approach.

  • Established a shared product vocabulary. I updated the entire product requirements document with consistent terminology and definitions, aligning product, design, engineering, and marketing around a common understanding of the experience.

Reconstructed original and revised publishing flows with direct sharing, deliberate public or private visibility, and creation notifications

IMPACT

  • Established a clearer creation and distribution model. Creators could share games directly with friends or complete the publishing flow with an explicit choice between public and private visibility.

  • Unblocked product implementation. By resolving the interaction and terminology questions, I gave the team a coherent mental model to build against.

03 / TERMINOLOGY

03 / TERMINOLOGY

Building shared language across platforms

Building shared language across platforms

Building shared language across platforms

The sharing and publishing work underscored to me the need to establish a clear terminology source of truth across content design for both the Eclipse mobile and web experiences (Horizon Create & Horizon Studio).

The sharing and publishing work underscored to me the need to establish a clear terminology source of truth across content design for both the Eclipse mobile and web experiences (Horizon Create & Horizon Studio).

The sharing and publishing work underscored to me the need to establish a clear terminology source of truth across content design for both the Eclipse mobile and web experiences (Horizon Create & Horizon Studio).

THE PROBLEM

Independent terminology decisions made across mobile and web threatened consistency as both products evolved.

MY APPROACH

  • Created a centralized terminology source of truth. I built a shared spreadsheet to document emerging terminology decisions, track unresolved questions, and surface inconsistencies across mobile and web.

  • Established a weekly CD alignment meeting. I created a recurring meeting with the content design team to evaluate terminology decisions against the needs of our distinct audiences (casual creators on mobile and more advanced creators on web) and determine where to align or differentiate.

IMPACT

  • Established intentional cross-platform consistency. I aligned terminology where concepts and interactions were shared, while preserving platform-specific language where different user needs warranted it.

04 / OUTCOME

04 / OUTCOME

What happened next

What happened next

What happened next

By the time I left the project, design and product leadership had approved a unified conversation strategy, engineering was incorporating it into the model, and we’d aligned on a clearer creation and publishing experience. I had also established a shared process for resolving terminology decisions across mobile and web.


I stepped away before external user testing began, so I don’t have performance metrics to share. The product was later publicly announced at Meta Connect 2026.

By the time I left the project, design and product leadership had approved a unified conversation strategy, engineering was incorporating it into the model, and we’d aligned on a clearer creation and publishing experience. I had also established a shared process for resolving terminology decisions across mobile and web.


I stepped away before external user testing began, so I don’t have performance metrics to share. The product was later publicly announced at Meta Connect 2026.

By the time I left the project, design and product leadership had approved a unified conversation strategy, engineering was incorporating it into the model, and we’d aligned on a clearer creation and publishing experience. I had also established a shared process for resolving terminology decisions across mobile and web.


I stepped away before external user testing began, so I don’t have performance metrics to share. The product was later publicly announced at Meta Connect 2026.