on-demand initiation

Gold definitionUpdated Apr 2, 2026

Definition

On-demand initiation is a design pattern for AI writing assistants where suggestions are provided only when explicitly requested by the user. This approach aims to empower writers by giving them control over AI intervention, thereby preserving their sense of authorship and agency.

At a glance

Executive summary

On-demand initiation is a design approach for AI writing tools where the AI only offers help, like sentence suggestions, when the user specifically asks for it. This method helps writers feel more in control and maintain their personal connection to their work, even while using AI to make writing easier.

TL;DR

It's a design for AI writing tools where the AI only helps when you specifically ask it to, so you still feel like the true author.

Key points

Core mechanism: User-explicitly-triggered AI assistance, such as sentence-level suggestions.
Problem solved: Weakened writer's sense of authorship and agency in AI co-writing.
Who uses it: Developers of AI writing assistants and human-computer interaction (HCI) researchers.
Vs main alternative: Proactive, always-on, or intrusive AI suggestions that may diminish user ownership.
Research trend: Designing human-centered AI for collaborative and creative tasks that prioritizes user agency.

Use cases

AI-powered email assistants that suggest rephrasing only when a user highlights text and clicks a 'rewrite' button.
Creative writing software where an AI offers plot twists or character descriptions only upon a specific user command.
Code completion tools that provide full function bodies only when explicitly invoked by a shortcut, rather than auto-completing aggressively.
Presentation software that suggests slide layouts or content only when the user clicks a 'design ideas' button.
AI art generators that offer style variations only after the user has defined initial parameters and explicitly requested alternatives.

Also known as

User-initiated AI, Opt-in AI assistance, Pull-based AI, Explicit AI guidance