Audit-IQ RegTech

Product Documentation

Last updated: March 2026

Audit-IQ helps organisations interpret regulatory documents and convert them into structured, structured, reviewable compliance outputs.

1. How the Platform Works

Step 1 — Upload Documents

Upload:

  • Regulatory frameworks
  • Internal policies
  • Client compliance documents

Supported formats:

  • PDF
  • Word (DOCX)
  • Web pages (public URLs)

Step 2 — AI-Assisted Extraction

Audit-IQ uses regulation-aware LLM prompts to:

  • Extract obligations and responsibilities from uploaded documents
  • Output structured, checklist-ready obligation lists
  • Run a keyword-and-reference coverage check against the regulation master list
  • Flag potential gaps for consultant review

Outputs are AI-generated and require human review before use in formal reporting.

Step 3 — Review & Export

You can:

  • Review extracted insights
  • Map evidence to obligations
  • Export data to PDF / Excel (beta)
  • Share summaries with teams or clients

2. Supported Frameworks

Frameworks with active AI extraction prompts and full regulation master lists:

  • APA 2024 (Australian Privacy Act — all 13 Australian Privacy Principles)
  • EU AI Act Title IV (Transparency obligations for certain AI systems)

Frameworks with checklist support (no AI extraction prompt — manual tracking only):

  • SOC 2 — Baseline (CC6 + CC7): logical access and system monitoring
  • ISO 27001 — 23 key controls across 9 domains
  • Essential Eight — 8 ACSC mitigation strategies (Beta)
  • ISO 31000 — 8 risk management controls (Beta)
  • AI Ethics — 8 AI governance controls (Beta)

3. Roadmap Frameworks

  • GDPR
  • NIS2
  • ISO 42001

4. Evidence Linking

When you select an evidence item, the platform scores it against your project's obligations and controls using keyword overlap and a built-in list of compliance-relevant terms. Results are ranked by score and displayed with a confidence level and the specific terms that drove the match.

How it works:

  • Suggestions are keyword-based — not LLM-generated
  • Confidence levels (high / medium / low) reflect keyword overlap strength, not a guarantee of relevance
  • Each suggestion shows the matched terms so you can judge fit before linking
  • Nothing is linked until you explicitly click Link — no auto-attachment
  • Control matches are shown for reference; links to controls are managed from the Controls tab
  • Adding a description to your evidence improves match quality

5. Known Limitations

  • All AI outputs require human review — do not use as sole basis for compliance decisions
  • Gap detection uses keyword and reference matching, not semantic understanding
  • PDF formatting (scanned documents, complex layouts) may reduce extraction quality
  • Only APA 2024 and EU AI Act Title IV have active extraction prompts today
  • Evidence linking uses keyword-based matching to surface suggestions — all links require a deliberate reviewer action; nothing is recorded automatically

6. Support

For documentation or product help:
📧 founder@audit-iq.com