Portfolio Case Study · Public Edition

Asset Studio — AI-Assisted Ecommerce Creative Operations

This page documents the product thinking, workflow design, creative tooling, and technical system behind Asset Studio. Credentials, customer data, private identifiers, proprietary prompts, provider payloads, and cloning instructions have been intentionally omitted.

My RoleProduct Owner · Creative Technologist · Builder
DisciplineCreative Operations · Internal Tools
StatusDeployed production application
DisclosureWorkflow and outcomes, sanitized
Creative Technology · Ecommerce · Production Systems

ASSET STUDIO

From collection truth to approved, reusable creative

Asset Studio brings collection selection, product sources, AI-assisted generation, finished-asset upload, visual editing, approval, metadata, reuse, and destination-draft preparation into one guided workspace.

Deployed production toolReact · OpenAI · AWSHuman approval required
01

The Problem

Collection creative moved between commerce pages, image sources, prompts, generators, design tools, folders, approvals, and campaign platforms. The context behind an asset—what it represented, how it was made, and whether it was ready—was easy to lose.

Inputs

Product truth lived elsewhere

Collections, product imagery, destinations, and merchandising context had to be rebuilt for every asset.

Creative

Generation was only one step

Useful output still required source selection, copy, layout, editing, dimensions, review, and approval.

Operations

Files lacked visible lineage

Clean sources, finished compositions, settings, usage history, and approval state were difficult to connect.

Delivery

Approval needed a controlled exit

Creative should move forward only when a human confirms the final asset and destination structure.

My contribution: workflow definition, product architecture, interface design, creative controls, React and backend implementation, AWS infrastructure, OpenAI integration, asset metadata, deployment, and operating documentation.
02

Product Architecture

Asset Studio is organized around four user-facing workspaces with a shared approval and storage model.

Asset StudioThe complete production workspace connecting commerce context, creative sources, generation, editing, approval, storage, and draft preparation.
CollectionsA guided builder for one to five email sections using live collection and product context.
DesignerA visual canvas for text, layout, crop, positioning, and approved exports.
Asset RepoA reusable library of saved sources, final assets, metadata, usage, and next actions.
COLLECTION → SOURCE → GENERATE OR UPLOAD → DESIGN → HUMAN APPROVAL → REPO + METADATA → DESTINATION DRAFT
03

One Guided Production Workspace

Collection creative

Open selector
Generate with AI
Upload finished asset
Use repo source
Three products
Collection queue
Example CollectionReady
Modern EssentialsNeeds asset
Weekend EditReview
Design ObjectsApproved
Selected asset
SANITIZED PREVIEW
Ready for reviewFinal approval →
Sanitized interface illustration. Names and states are fictional; no operational data is reproduced.
04

Creative Control After Generation

Asset Studio treats generation as an input to a creative workflow—not as the finished decision.

Sources

Choose the right visual role

Use collection heroes, product images, uploads, prior repo assets, references, or a defined three-product composition.

Canvas

Keep the work editable

Text, layout, image positioning, scale, crop, and output format remain under human control.

Asset integrity

Preserve clean sources

Generated source imagery is stored separately from text-composited and finished outputs.

Approval

Human review is the release gate

An asset is not treated as final until a user explicitly approves the current version.

Design principle: automate the repeatable setup and production work while keeping creative judgment, corrections, and final release visible to the user.
05

Asset Repo and Provenance

The Asset Repo makes approved work reusable without separating the file from the information needed to understand it later.

Metadata

Sidecar records

Saved assets retain structured provenance, workflow settings, source roles, and relevant collection context.

Reuse

Start from prior work

Users can open an asset in design mode or rerun a workflow with logged settings.

Operations

Manage the library

Filter, inspect metadata, download, select, delete, and analyze approved asset groups.

Usage

See recent destination history

Recently used assets are visibly marked so the team can avoid accidental repetition.

06

System and Delivery Boundary

The product combines a focused frontend with serverless application services and approval-gated movement toward a destination draft.

LayerRolePublic description
React + TypeScriptCreative workspaceCollections, Bulk Create, Designer, Asset Only, Asset Repo, and Settings.
Express + LambdaApplication servicesGeneration jobs, collection context, asset operations, workflow records, and draft preparation.
S3 + DynamoDBAssets and stateApproved files, metadata sidecars, collection workflow status, and configuration records.
OpenAICreative assistanceImage and copy assistance inside an editable, human-reviewed process.
Destination integrationDraft handoffApproved sections can be assembled into a draft when the destination service is configured.
Public preview: a limited, unlinked preview provides Asset Studio without the surrounding internal navigation and caps image generations per visitor. Full operational access remains restricted.
07

What Exists Now

CapabilityStatusPublic description
Collections builderDeployedOne to five sections using collection, product, upload, and repo sources.
Bulk CreateDeployedQueue multiple collections, formats, and category treatments for review.
DesignerDeployedEditable composition, text, placement, sizing, and export controls.
Finished-asset uploadDeployedJPG, PNG, and GIF intake without forced copy generation.
Asset Repo + metadataDeployedReusable asset library with sidecars, analysis, usage, and bulk actions.
Public previewLimitedExternal testing surface with restricted scope and generation limits.
Automated image evaluatorRemovedHuman approval remains the active release decision; provenance metadata is retained.
08

Public / Private Boundary

Safe to show

  • Problem, workflow, and my role
  • Collections / Designer / Repo model
  • Sanitized interface structure
  • Human approval and provenance design
  • High-level serverless architecture
  • Honest implementation status
  • Limited public preview

Intentionally withheld

  • Credentials and private identifiers
  • Customer, supplier, and internal records
  • Proprietary prompts and generation recipes
  • Provider payloads and exact API sequences
  • Private schemas and business rules
  • Source files and production resource names
  • Details sufficient to clone the system
Portfolio principle: demonstrate product judgment, creative operations, technical execution, and validation without publishing the operational blueprint.

Try the limited public preview.
The preview is isolated from internal navigation and operational data.

Launch preview ↗