01 / AI interaction design

Online alpha

StateLab

Design what happens between the prompt and the result.

An interaction lab that turns AI content-generation rules into working prototypes of waiting, approval, failure, and recovery—so teams can experience behavior before building it.

StateLab conceptual illustrationWAITCONFIRMRETRYRECOVERCONCEPT ILLUSTRATION · NOT AN APPLICATION SCREENSHOT

THE DESIGN QUESTION

How can a team experience the difficult moments of an AI workflow before those moments reach users?

Make behavior rules explicit, generate a three-step simulation, and let a designer experience six scenarios while recording their own observations and decisions.

My contribution
Product adaptation, behavior-rule design, six scenario simulations, human review and export flows, and the controls around optional AI editing.
Current scope
Online alpha · browser-based simulation
Exploring
Agentic UX · Scenario testing · Harness

THE WORKFLOW

01

Describe

Define the content-generation feature or choose a template.

02

Decide

Confirm cancellation, partial-result preservation, and retry approval rules.

03

Rehearse

Generate a template-based prototype and trigger six scenarios.

04

Review

Record human checks, then export HTML, a handoff note, or a JSON backup.

A CLOSER LOOK

One feature. Four moments to design.

Illustrative walkthrough · the interactive alpha is available above

01 / 04

A video is still generating.

Show meaningful progress and a clear next action, without promising an unverified completion time.

These examples explain the interaction rules. The online alpha simulates outcomes and credits; it does not generate real content or perform an automated UX evaluation.

UNDER THE SURFACE

Make the process legible.
Make the boundaries explicit.

Agent · optional

The local application includes structured AI planning and a code-editing agent limited to 12 tool steps. Model calls are disabled in this public alpha; its templates and scenarios work without AI.

Harness · implemented

Confirmed rules, six scenario inputs, a restricted prototype file, and an isolated iframe with network restrictions make execution boundaries explicit. Changes invalidate earlier human review results.

Verification · human-led

A static checker verifies required prototype interfaces. Reviewers operate each scenario and record their conclusions; these checks do not constitute automated browser evaluation.

TODAY

Online alpha

The online alpha supports templates, six simulated scenarios, manual review, browser-local saving, and HTML/Markdown/JSON export. It performs no real generation or payment.

WHAT’S NEXT

Still exploring

Evaluate AI-assisted editing against the scenario rules, then expand beyond the fixed three-step content-generation template.

Adapted from a downloaded UIGen starter. The StateLab product layer adds scenario rules, simulation, review, persistence, and handoff. The public release is a browser-only build with AI features disabled.