Artifact note

This accessible HTML version preserves the structure and content of the original project artifact. The downloadable PDF remains available as the source document.

Core principle

Scaffold is designed as a support system rather than a productivity optimization engine.

The system intentionally prioritizes

  • autonomy
  • reversibility
  • transparency
  • reduced emotional escalation
  • bounded scope
  • informed consent

Over

  • surveillance
  • behavioral manipulation
  • compliance optimization
  • coercive automation
  • engagement maximization

Scaffold explores how support systems might reduce cognitive burden while remaining transparent, interruptible, optional, and consent-directed.

What this system is — and is not

Scaffold is

  • A consent-based assistant for sequencing, reminders, and transitions
  • Designed to support users during overwhelm, executive dysfunction, or cognitive overload
  • Built to reduce cognitive friction rather than maximize output
  • Designed to remain interruptible, inspectable, and optional

Scaffold is not

  • A productivity optimizer
  • A behavior-shaping system
  • A replacement for human judgment
  • A system that acts without permission

If a task is skipped, postponed, or abandoned, the system does not escalate or attempt to correct the user.

Ethical design constraints

No action without explicit approval

The system may suggest actions, but it may not execute actions without explicit user approval. There are no silent automations or irreversible actions without confirmation.

Closed by default

By default, the system:

  • does not browse the open internet
  • does not retrieve third-party data
  • does not access external systems
  • does not expand scope without user opt-in

No pooled behavioral learning

The system does not:

  • aggregate behavioral profiles across users
  • blend user histories
  • learn from other users’ conversations
  • transfer personal context between accounts

User information remains scoped to the individual interaction environment.

Transparency is required

Users can inspect:

  • what information informed a response
  • what actions were proposed
  • what data is stored
  • why particular actions were suggested

If the system cannot clearly explain its reasoning, it should not act.

No psychological pressure loops

The system avoids:

  • streak mechanics
  • guilt framing
  • escalating reminders
  • urgency loops
  • compliance tracking
  • manipulative notifications

Missed tasks are treated as neutral signals that inform future interactions rather than failures.

Interaction constraints

Interaction design

  • Clear “yes / no / not now” controls
  • Immediate pause or disengagement available
  • Preference for reversible actions
  • Limited-scope interactions over persistent escalation

Executive function support

  • Break tasks into smaller actionable steps
  • Reduce ambiguity during task initiation
  • Support transition management
  • Adjust interaction pacing based on user-stated capacity

Logging & control

  • Action history remains visible
  • Stored information can be exported or deleted
  • The system can be disengaged without penalty

Design philosophy

The project evolved away from asking:

“How can systems make users more productive?”

Toward:

“How can systems reduce cognitive friction while preserving agency, reversibility, and supporting the user’s emotional regulation?”

Operational relevance

The interaction constraints explored in Scaffold increasingly apply to:

  • AI-supported workflows
  • accessibility systems
  • knowledge-work platforms
  • educational technologies
  • neurodivergent user experiences
  • human-AI interaction systems