Overview
How might we help people start recurring AI jobs with a proven prompt shape without locking them into a rigid macro?
When the same job repeats
- Writing, marketing, support, and critique tools where people run the same brief weekly.
- Composers that expose a template library or tool-scoped starters (ChatGPT Image, Notion AI, Jasper).
- Products that teach good prompt structure with slots people fill before send.
- Teams that want shared, editable starters instead of tribal prompt docs.
When every turn is unique
- Highly exploratory chat where every turn is unique and a template library would go stale.
- Expert power users who already have personal prompt libraries outside the product.
- Products with one fixed prompt shape where “templates” are just duplicate empty states.
States
Design pick → fill → edit → send, not only a list of canned strings.
Browsing
A library or tool-scoped set of templates is visible. Categories or search help when the list grows.
Selected
Choosing a template drops its structure into the composer. Slots and fixed copy are both visible.
Filling
The person completes variables (audience, goal, constraints) without leaving the input.
Editing
They can rewrite any part before send. The template is a draft, not a sealed script.
Sent
The customized prompt goes as a normal user turn. History shows what was actually sent.
Reusable
Optional save or pin keeps a refined version in the library for the next run.
Key UX elements
The parts that must be present for templates to accelerate work without becoming brittle macros.
Show a scannable set of jobs.
Cards or rows with clear labels beat a dump of full prompts. Group by category when the list grows.
Mark what the person must fill.
Placeholders for audience, goal, or input make the template customizable instead of one-size-fits-all.
Drop the structure into the composer.
Selecting a template should land editable text in the input, not auto-send or open a dead-end modal.
Keep every character changeable.
Locked templates teach nothing and fail real context. Edit-before-send is the default.
Swap the set when the tool or mode changes.
Image templates should not appear in search mode. Tie the library to the active capability.
Let people keep a refined version.
Pin or save after a good run so the library improves with use, not only with shipped defaults.
Anti-patterns to avoid
Templates that cannot be edited before send, forcing a rigid phrasing.
No categories or search in a long template list.
Templates that invoke tools or spend credits with no preview of side effects.
Stale templates that no longer match current model or product capabilities.
How products use it
| Product | Implementation |
|---|---|
| ChatGPT | Custom GPT instructions and example starters act as reusable template patterns. |
| Notion AI | Writing and meeting templates drop structured prompts into the page. |
| Claude Projects | Project instructions and canned prompts standardize recurring workflows. |
| Jasper / marketing AI | Campaign-type templates with fill-in fields for audience and offer. |
Real-world examples
How shipped products implement prompt templates, from our teardown guides.
ChatGPTChatGPT composer UX: input bar, tools & voice design
Design the composer
ClaudeClaude composer UX: model picker, skills & attachments
Design the composer
GeminiGemini composer UX: tools, Drive & thinking modes
Design the composer
Google AI ModeGoogle AI Mode composer UX: Canvas, Create & Lens
Design the composer
GrokGrok composer UX: skills, models & voice input
Design the composer
LovableLovable composer UX: Build, Plan & templates
Design the composer
ManusManus composer UX: assign tasks, modes & agent tiers
Design the composer
ManusManus onboarding UX: role segmentation & Pro upsell
Getting to first value
PerplexityPerplexity composer UX: search modes & connectors
Design the composer
PerplexityPerplexity personalization UX: vertical profiles, response style & memory
Personalization & memory
Implementation
Copy this prompt to generate a production-ready implementation in Cursor, Claude Code, Lovable, or any AI coding agent.
Generate a production-ready implementation of the "Prompt Templates" AI interface design pattern.
Pattern Definition:Frequently asked questions
What are prompt templates in AI UX?
Prompt templates are reusable, often fill-in-the-blank starters that help users begin a high-quality AI turn without inventing the full prompt from scratch.
How are templates different from empty-state prompt starters?
Empty-state starters are a few nudges on first open. Templates are a maintained library users return to for recurring jobs, usually with slots, categories, and save/share.
Should templates be editable before send?
Yes. Editable templates teach prompting and fit the user’s context. Locking the text turns templates into brittle macros.