Semantic Zoom
Complexity on a Single Axis
The same AI output rendered at three levels of abstraction — Headline, Summary, Full Detail — controlled by a single slider. Users navigate complexity without losing context, zooming in only when they need more depth.
Framing
The problem
One output depth rarely fits both skim and deep work.
The pattern
Bind one answer to multiple abstraction levels on a single control.
Why chat breaks here
Users must repeatedly ask for shorter or longer versions and lose continuity.
Risks
Compression levels can hide nuance or imply false completeness.
Avoid when
Meaning does not survive well across summary and detail layers.
Use when
One answer needs to serve both skim and deep-read modes without rerunning generation.
DOPE evaluation
- Directability
- One slider (or a click on any level label) moves through all three representations in both directions — no regeneration, no lost position
- Observability
- A persistent indicator states the position exactly — "Level 2 of 3 — Summary · ≈ 58 words · 0.5% of source" — so depth is read off the control, not remembered from the conversation
- Predictability
- Each level has a declared contract — 1 sentence, 4 key findings, 3 cited sections — so the reader knows what a zoom step returns before dragging
- Explainability
- Zooming changes the representation, not the crop — sentence to labeled findings to cited prose — and each level names its method and exact word count
In the wild
- NotebookLM Audio Overview (Google) — Length control: Shorter / Default / Longer for audio overviews, plus formats (Brief / Deep Dive / Critique / Debate). One-control-multiple-depths, exactly the pattern.
- Granola (Granola) — After a meeting, switch between TL;DR / Full / Action Items tabs on the same source content — semantic zoom as tab switcher.
- Apple Intelligence Summaries (Apple) — Caveat — closer to format presets (Summary / Key Points / List / Table) than abstraction-depth on a single axis. Still useful as a zoom-adjacent shipping reference.
FAQ
When should I use the Semantic Zoom pattern?
One answer needs to serve both skim and deep-read modes without rerunning generation.
When should I avoid the Semantic Zoom pattern?
Meaning does not survive well across summary and detail layers.
What problem does Semantic Zoom solve?
One output depth rarely fits both skim and deep work.
Why is chat the wrong fit for this?
Users must repeatedly ask for shorter or longer versions and lose continuity.
Related patterns
- Alternative to: Progressive Disclosure — Output layers along one slider vs control layers along a complexity stack.
- Often paired with: Reasoning Trace — A reasoning chain becomes navigable across abstraction depth.
- Often paired with: Citation Trail — Zoom in on a claim and the contributing sources surface in place.