Skip to content
Scallar

Trust

Understand work. Respect people.

How we intend Scallar to behave — and an honest account of where we are today.

Status — please read

Scallar is in a pre-launch phase. The principles below describe intended practices guiding product design. They are not certifications or audited compliance statements. Formal policies (security, data processing, privacy) will be published here once approved.

  1. 01

    Privacy by design

    Individual detail is visible to the individual concerned. Visibility boundaries are part of the architecture, not a setting.

    Intended practice
  2. 02

    Appropriate aggregation

    Managers work with team-level information. No drill-down into personal AI usage histories.

    Intended practice
  3. 03

    Human oversight

    Consequential decisions require validation by an accountable person. Evaluation and compensation stay at Levels 1–2.

    Intended practice
  4. 04

    Explainable recommendations

    Every recommendation should show which signals informed it and what remains uncertain.

    Intended practice
  5. 05

    Contestability

    People can contest a recommendation or conclusion that concerns them.

    Intended practice
  6. 06

    Responsible data use

    Only authorized sources, used for the stated purpose. Organizational intelligence, not surveillance.

    Intended practice
  7. 07

    Regulatory considerations

    Designed with the EU AI Act's treatment of workplace AI in mind. No certification or audit is claimed.

    Intended practice

Individual view vs manager view.

Individual detail is visible to the individual concerned. Managers see aggregated team-level information, with no drill-down. Aggregation alone doesn't guarantee anonymity in small teams, so visibility thresholds are part of the design.

Visible only to you · Sam (fictional)

  • Focus time this week6h 20m
  • Out-of-hours activity↑ 3 evenings
  • AI-assisted contributions14
  • Private suggestion: block two focus mornings next week.

Four levels of autonomy.

  1. Level 1

    Private information

    Inform the individual directly. Only they see it.

  2. Level 2

    Manager recommendation

    Recommend an action to an authorized human.

  3. Level 3

    Execution with approval

    Execute an authorized action after validation.

  4. Level 4

    Autonomous low-risk

    Execute permitted low-risk actions autonomously.

Hard rule: anything relating to evaluation or compensation is restricted to Levels 1–2. Scallar never decides salaries, promotions, ratings, disciplinary actions or dismissals.

Established policies

Security overview — pending company approval

Data processing terms — pending company approval

Sub-processors — pending company approval

Questions about our approach? We'd like to hear them.

Talk to the team