Blueprint · product band

Participation controls

Scraping and Local AI are explicit, truthful controls: readiness, consent, background limits, and withdrawal stay visible.

Cell 5.1 · product · privacy · runtime
  • Scraping · Local AI pills to Consent + readiness: open explanation
  • Consent + readiness to Truthful device state: observe capability
open explanationobserve capability
perspectivemixed

Scraping · Local AI pills

Persistent participation state in the product band.

CareerVector UI
policytarget

Consent + readiness

Explain download, hardware, browser, foreground/background, and withdrawal.

Workspace participant
processtarget

Truthful device state

Disabled, preparing, ready, active, limited, failed, or unavailable.

CareerVector device
participation statecapability supports work
  1. 01 · perspective · mixed Scraping · Local AI pills

    Persistent participation state in the product band.

    CareerVector UI
  2. 02 · policy · target Consent + readiness

    Explain download, hardware, browser, foreground/background, and withdrawal.

    Workspace participant
  3. 03 · process · target Truthful device state

    Disabled, preparing, ready, active, limited, failed, or unavailable.

    CareerVector device

Flows

  • Scraping · Local AI pills Consent + readiness open explanation
  • Consent + readiness Truthful device state observe capability
as built target / open Boundary arrows open the neighbouring module; ports stay machine-only

CareerVector Experience

Purpose

Explain the complete interaction model and visual language of CareerVector. The experience spans private product work, optional public contribution, AI execution, support, and system state; it cannot be represented by the central product square alone.

Owns

  • Workspace, radar, CV/CL, job, evaluation, application, and collaboration interactions.
  • Consistent behavior across browser and Tauri desktop delivery.
  • Human UI, API, and MCP perspectives over the same canonical workspace state.
  • Scraping and Local AI pills as truthful, consent-based participation controls.
  • “Smart Routing AI” explanations, explicit overrides, BYOK, and local/included availability.
  • Visual language for origin, activity, uncertainty, freshness, failure, and recovery.

Does not own

  • Canonical public role/ad evidence.
  • Scraper scheduling or routing mathematics.
  • Provider policy, legal interpretation, or operator authority.

Current state

The shared application shell, collaborative workspace, web and desktop participant bridge, Scraping pill, adaptive AI routing foundation, and explanation surfaces exist at different levels of completeness. The Local AI adoption experience and several product workflows remain target work.

Agent contract

User-visible changes require browser evidence. Shared UI is extracted only where behavior and semantics are truly shared; JobCache and CareerVector are not skins over one product.

Completion evidence

Complete responsive journeys in web and desktop; accessibility; offline and collaboration behavior; visual snapshots; route explanations; consent and withdrawal; and error recovery.

Source: architecture/modules/careervector-experience.md