Origination Prompt

This page documents the master prompt and research context that generated 10 autonomous guided checklist variants, each exploring a fundamentally different approach to surfacing and resolving cross-page configuration issues in enterprise software.

Verbatim Origination Prompt
I want to explore multiple versions of an idea for a "bolt-on" guided checklists concept.

Exposing the guided checklists could include: floating action button, notification/checklist icon in menu bar with warnings, on-page widgets, in main menu.

The checklists could be: modal, action sheet, drawer, side-panel, full-page.

The checklists could be: programmatic (this problem is occuring and to fix it you need to click this link and do this thing, this problem is occuring so we analyzed it and here are some options (a, b, ...n), this problem is occuring and we need to investigate it more together to see what you think the primary cause is and address it).

We should have concepts for the ai generated options with "helpful/not helpful" which could be represented a number of different ways.

The idea is that lots of enterprise software is hard to reason about and frequently problems on one-page are really cuased by configs, rules, states on other pages and there isn't always a clear path to remediation.

Consider this from both a mobile & web perspective BUT also important consider this from multiple user personas.

1. Consider all the permutations that would be interesting. Not just what I listed.
2. Build the prompt for each of those versions.
3. Then spawn of sub-agents to build each of those versions.
4. Then update the changelog (with this origination prompt) and each version that gets added.
5. Then update the experiments.html.

Research Domains

Enterprise UX Patterns

Error states, configuration wizards, help systems, tooltips, guided tours, in-app messaging — how enterprise software currently handles complexity.

Cross-Page Dependencies

Settings inheritance, rule cascades, permission hierarchies, feature flags, integration configs — the hidden connections that cause "spooky action at a distance."

Mobile Enterprise Patterns

Bottom sheets, action sheets, thumb zones, swipe gestures, notification badges — touch-friendly patterns for complex workflows.

AI/ML Feedback Loops

Helpful/not helpful, confidence scores, explanation traces, correction mechanisms — how to make AI suggestions improvable.

User Persona Spectrum

Power admins, new users, executives, support agents, mobile field workers — different mental models and interaction preferences.

Intervention Timing

Proactive vs reactive, predictive warnings, auto-remediation, investigation collaboration — when and how to surface issues.

The 10 Variants

01
Command Center
Side panel with keyboard shortcuts — power admin interface for batch issue management
"j/k to navigate, x to resolve, / to filter."
web power user
02
Gentle Guide
Stepped modal wizard — educational onboarding with celebration states
"Let me show you why this matters."
web new user
03
Mobile Fixer
Bottom sheet with one-tap fixes — thumb-zone optimized for field workers
"Fix it now, you're in the field."
mobile casual user
04
AI Collaborator
Conversational chat drawer — collaborative diagnosis with reasoning traces
"Let's figure this out together."
both ai feedback
05
Executive Dashboard
Full-page health view — strategic overview with delegation and confidence scores
"Your system health at a glance."
web executive ai feedback
06
Contextual Whisper
Inline indicators with popovers — non-intrusive hints discovered while working
"This field is affected by settings elsewhere."
web in-flow
07
Support Toolkit
Split-pane with customer context — support agent view with history and scripts
"Here's what's wrong with their account."
web support agent
08
Proactive Monitor
Notification drawer with predictions — early warnings before problems occur
"This will become a problem in 3 days."
web ops admin ai feedback
09
Task Triage
Swipeable card stack — gamified batch processing with streak indicators
"Swipe right to fix, left to defer."
mobile power user
10
Ambient Intelligence
Auto-remediation with undo toasts — minimal friction for interruption-averse users
"We fixed it. Undo if we got it wrong."
both ai feedback

Coverage Matrix

Dimension Coverage
Exposure FAB (v1), beacon widgets (v2), tab bar (v3), chat bubble (v4), main menu (v5), inline indicators (v6), toolbar (v7), notification bell (v8), card stack (v9), status bar (v10)
Presentation Side panel (v1), modal wizard (v2), bottom sheet (v3), chat drawer (v4), full-page (v5), popover (v6), split-pane (v7), notification drawer (v8), card stack (v9), toast/snackbar (v10)
Problem Type Batch actions (v1), educational (v2), one-tap fix (v3), investigation (v4), delegation (v5), contextual hints (v6), customer-specific (v7), predictive (v8), triage queue (v9), auto-fix (v10)
Persona Power admin (v1), new user (v2), mobile field worker (v3), any/complex (v4), executive (v5), regular user (v6), support agent (v7), ops admin (v8), mobile power (v9), interruption-averse (v10)
AI Feedback Heavy (v4, v10), Medium (v5, v8), None (v1, v2, v3, v6, v7, v9)

Execution Model

1 Research Agent — Analyzes exposure methods, presentations, problem types, personas, platforms
↓
2 10 Build Agents (parallel) — Each agent receives a detailed prompt and builds one variant autonomously
↓
3 Validation Agent — Updates experiments.html, changelog.html, verifies all links and patterns

Design Constraints

Every variant follows the UI Explorations framework rules: pure HTML/CSS, no JavaScript, no external dependencies beyond Google Fonts. Each includes the standard experiment sidebar with origination context, a device preview switcher for desktop/mobile/fullscreen views, and realistic enterprise sample data.