> ## Documentation Index
> Fetch the complete documentation index at: https://docs.hr-easy.nlead.ch/llms.txt
> Use this file to discover all available pages before exploring further.

# Design philosophy

> The Compassionate Systems Framework — and the things this product deliberately will not build.

Swiss HR Easy is filtered through the **Compassionate Systems Framework**
([ADR-006](/development/adr)). This is not decoration on a feature list; it decides
what gets built and what gets refused.

The framework has three domains, and the modules map onto them:

<CardGroup cols={3}>
  <Card title="Self-awareness" icon="user">
    Reflective check-ins, strengths reflection, growth journeys. The person is the audience for
    their own data.
  </Card>

  <Card title="Each other" icon="handshake">
    Feedforward, peer recognition, growth dialogues, the advice process. Feedback is a conversation,
    not a verdict.
  </Card>

  <Card title="System" icon="diagram-project">
    Team vitality sensing, 360° perspective gathering, decision methods that distribute authority.
    Decision-making reveals an organization's real power structure — so make it visible.
  </Card>
</CardGroup>

## What the product will not do

These are hard product constraints. A ticket asking for one of them gets referred back
to the ADR rather than estimated.

<AccordionGroup>
  <Accordion title="No forced ranking, bell curves, or stack ranking" icon="ban">
    Rating schemes exist and are expressive, but the schema deliberately does not surface top-10% /
    bottom-10% reports. There is no path from "rate your team" to "here is who to cut".
  </Accordion>

  <Accordion title="No performance-improvement-plan defaults" icon="ban">
    Probation and feedback flows are framed as decisions about **fit and support**, not as legal
    cover for a termination that has already been decided. See
    [Probation](/guides/lifecycle/probation).
  </Accordion>

  <Accordion title="No gamification of HR processes" icon="ban">
    No leaderboards, no streaks, no badges, no scores presented as individual rankings.
  </Accordion>

  <Accordion title="No dark-pattern retention nudges" icon="ban">
    Offboarding is an **exit dialogue**, not an exit interview designed to talk someone out of
    leaving. Cancellation is always one click away.
  </Accordion>

  <Accordion title="No pseudo-objective AI scores presented as facts" icon="ban">
    Every AI ranking or observation must show its criteria and the data it used, and the human keeps
    the decision. See [AI transparency](/guides/ai/transparency).
  </Accordion>
</AccordionGroup>

## How this shows up concretely

**Feedback** has no 1–5 scale and no forced distribution. Cycles collect narrative and
reflective responses, and continuous feedback is peer-to-peer feedforward rather than
manager-to-report scoring.

**Surveys** treat anonymity as a technical property, not a promise: an anonymous
response never persists the person→answer link at all — the invitation records only
*that* someone responded. Results, the AI report and the export stay locked until a
configurable `minResponsesForResults` threshold is met, so a small audience cannot be
re-identified by elimination.

**Compensation** explains pay policy — roles, bands, and where a person sits in one and
why. The employee self-view never compares them against a colleague.

**Salary visibility** is closed by default. Even HR staff read salaries only through
hand-picked grants; managers read strictly below themselves and never at or above their
own level.

**Probation** is a shared meeting template used *together* by HR, the employee and
their manager, with reflective questionnaires and no scores.

**Decisions** offer consent (sociocracy), the advice process, systemic consensus with
resistance rating, dot voting and delegation poker — methods whose point is to
distribute authority rather than to record a manager's call.

<Note>
  If you are extending the platform and a requirement collides with this page, raise it in review
  and link here. The framework wins.
</Note>