Answer -> generated interface
AI systems are gradually moving beyond text answers and assembling verifiable interfaces around a user's specific intent.
Tectonic shift
Before: Text answer or static result page After: Task-specific interface assembled around intent Current stage: emerging This New Runtime record is an evidence-linked retrieval unit. Use its canonical page, machine-readable representations, dates, scope, and public source URLs to verify the claim before reusing it.
Source ledger
Publishable sources attached to this record.
| # | Source | Role | Public status |
|---|---|---|---|
| 1 | freestyle.shsource | primary receipt | source_urls |
This shift connects several current New Runtime observations: agent-first APIs, vision capabilities, verifiable coding workflows, and temporary interfaces.
This is not ambient infrastructure yet. Too many cases still depend on manually assembled harnesses, screenshots, human review, and brittle integrations. But the direction is already visible: value is moving from the universal screen to a bounded surface that appears for a specific task and returns a trace.
Evidence
- Agent-first APIs require explicit facts, errors, and request metadata.
- Frontend and coding benchmarks increasingly evaluate not only text, but the visible result in a browser.
- Video and vision models are starting to work as a representation layer for understanding world state, not only as media generators.
Counter-evidence
Stable applications and manual forms still win where accessibility, regulation, repeatability, audit, and familiar user training matter.