The First New UI Paradigm
in 60 Years
A shift from Command-Based ("Do this, click that") to Intent-Based ("Here is the result I want") interaction.
The Affordance Gap
AI creates a new usability crisis: users do not know what they can ask for.
Faster Execution
Speed gain for info retrieval using GenUI vs click paths.
Error Risk
Increase when users over-rely on AI without verification.
Threshold
If prompting takes longer than clicking, users revert to GUI.
5 Core Shifts
Malleable Software
UI generated at runtime based on intent. The interface is the response.
Sandwich Workflow
Human sets context → AI generates → Human verifies. Users reject full automation.
Inverse IA
Start with the destination (via intent), system reconstructs the navigation path.
Latency Trade-off
If AI >1s latency, it must provide >10x value. AI cannot replace nav until instant.
Contextual Chips
Dynamic buttons that suggest intents based on context. Bridges Recognition (GUI) and Recall (AI).
Smart Apps
Add an intelligent layer on top of existing apps. No rebuild required.
The Stack
- ✓ App acts as an agent
- ✓ Complex workflows become conversation
- ✓ Preserve existing investment
FunctionGemma
270M Parameters
- On-Device: Zero latency, runs on phone/laptop
- ~300MB model file, ~500MB RAM
- ~85% task success rate
- GDPR Ready: Data never leaves device
Data Sovereignty
Data never leaves browser or private network
Offline Capable
Works anywhere, <200ms latency
Cost Advantage
Offload compute to edge, reduce cloud costs
Implementation Matrix
"Go to Settings." AI is overkill for 200ms actions.
"How do I change privacy?" AI suggests the path.
"Summarize Q3 reports and email them." AI is 10x faster.
Hybrid Interfaces Win
Neither pure GUI nor pure AI. The future is intelligent augmentation of existing applications with context-aware, intent-based interactions.
Read Full Research →