Skip to main content

User Guide

In a nutshell

What this document is: A guide for end users reviewing Clawscan outputs.

Why this matters: Ensures that detected risks are interpreted correctly and handled through appropriate internal processes.

When to use this: During day-to-day compliance monitoring and review activities.


Overview

Clawscan identifies communications that may present legal or compliance risks.

Users interact with:

  • flagged communications
  • risk indicators
  • structured explanations

The system supports users in identifying relevant signals but does not replace human judgement.


What users see

Clawscan provides risk signals associated with communications.

These may include:

  • a risk classification (e.g. competition law, anti-corruption)
  • a numerical risk score
  • a reasoning summary explaining the detection

These elements help prioritize which communications should be reviewed.


How to interpret results

Clawscan outputs should be understood as:

  • indicators, not conclusions
  • prioritisation tools, not decisions
  • support for review, not replacement of expertise

A flagged communication does not necessarily indicate a violation.


Review workflow

A typical workflow includes:

  1. Review the flagged communication
  2. Assess context and intent
  3. Determine relevance and risk level
  4. Escalate if necessary

Organizations should define their own internal processes for handling alerts.


False positives and detection strategy

Clawscan prioritizes early detection.

This means:

  • some false positives may occur
  • missing a risk is considered more critical than over-flagging

Users should:

  • apply judgement during review
  • rely on internal expertise
  • avoid automated conclusions

Human oversight

Clawscan does not make automated decisions.

Users remain responsible for:

  • interpreting signals
  • making legal assessments
  • deciding on actions

This ensures:

  • proportionality
  • accountability
  • compliance with regulatory expectations

See:


What Clawscan does not do

Clawscan:

  • does not evaluate employees
  • does not create behavioural profiles
  • does not generate individual performance metrics
  • does not make disciplinary recommendations

The system is designed to analyze content, not individuals.


Good practices

Organizations are encouraged to:

  • define clear review procedures
  • ensure appropriate access controls
  • provide training to reviewers
  • document decisions and actions

Data visibility

Users only access:

  • communications flagged for review
  • associated risk indicators

Access should be restricted to authorized personnel.

See: