ReqTangle uses artificial intelligence to help you write better software requirements. This page explains how our AI features work, what data they use, and their limitations. Our AI assists your work, it doesn't replace your professional judgment.

​​

Key Points:

• AI suggests improvements to your requirements

• All processing happens in the EU (Sweden)

• Your data is never used to train AI models

• You control when and how AI is used

• AI outputs may contain errors, always review before using

​

1. Why AI in Requirements Management?

​

Writing clear, complete, and consistent requirements is challenging. ReqTangle uses AI to:

• Analyze requirement quality - Check for ambiguity, completeness, and clarity

• Suggest improvements - Offer alternative phrasing and missing details

• Identify dependencies - Spot relationships between requirements

• Enhance consistency - Flag terminology and style inconsistencies

​

Important: AI is an assistant, not an expert. You remain responsible for the accuracy and appropriateness of your requirements.

​

2. How ReqTangle's AI Works

​

What Triggers AI Processing

​

AI features activate only when you explicitly use them:

​

✓ Clicking "Analyze Requirement" or "Improve Quality" buttons  
✓ Using auto-completion suggestions  
✓ Running dependency analysis on requirements  
✓ Requesting AI-powered summaries or explanations

 

AI does not run automatically in the background. You control when your data is sent for AI processing.

 

What Data is Sent to AI Systems

When you use AI features, ReqTangle sends:

​

- The text of the specific requirement(s) you're working with
- Relevant context (related requirements, if you enable this)
- Your feedback on AI suggestions (to improve the experience)

​

What we DO NOT send:

- Your personal account information
- Requirements you haven't chosen to analyze
- Data from other users or projects
- Any sensitive metadata about your organization

​

Where AI Processing Happens

All AI processing occurs within the European Union:

- AI Provider: Codicent Inside AB (Swedish company)
- Infrastructure: Microsoft Azure Sweden Central (Sweden Central datacenters)
- Data Residency: 100% EU-based, no data transfers to third countries

​

Processing Duration

AI processing is temporary:

1. Your requirement text is sent to Codicent's AI engine
2. The AI analyzes it and generates suggestions
3. Results are returned to you immediately
4. Processing data is not retained beyond your session

The AI engine does not permanently store your prompts or requirements.

​​

3. AI Technology & Models

 

Underlying Technology

ReqTangle's AI features are powered by:

- Large Language Models (LLMs) - Advanced natural language processing systems
- Specialized fine-tuning - Optimized for requirements engineering terminology
- Retrieval-augmented generation - Contextual understanding of software development

 

Provider: Codicent Inside AB licenses and operates the AI technology on our behalf.

 

Model Limitations

AI language models have inherent limitations:

⚠️ Probabilistic Outputs - AI generates statistically likely responses, not guaranteed correct answers 
⚠️ No Guarantees of Accuracy - Suggestions may be incomplete, misleading, or incorrect  
⚠️ Context Limitations - AI may misinterpret domain-specific or technical terminology  
⚠️ Bias Risks - AI models can reflect biases present in their training data  
⚠️ Non-Deterministic - The same input may produce different outputs at different times  

​

Your responsibility: Always review and validate AI suggestions before incorporating them into your work.

​

4. Your Data & AI Training

​

Your Data is Never Used for Training

 

We explicitly guarantee:

✓ Your requirements data is never used to train AI models  
✓ Your data is never shared with other ReqTangle users  
✓ Your data is never sold or licensed to third parties  
✓ Your data is never used to improve AI for other customers  

​

How AI Models are Trained

The underlying AI models are trained on:

- Publicly available text from the internet
- Open-source software documentation
- Academic papers and requirements engineering literature
- Synthetic and anonymized datasets

 

Your private requirements remain private.

 

5. Human Oversight & Control

 

You're Always in Control

ReqTangle's AI follows the principle of meaningful human oversight (EU AI Act Article 14):

✓ You decide when to use AI features  
✓ You review all AI suggestions before accepting them  
✓ You can edit or reject any AI output  

 

AI Never Acts Autonomously

AI in ReqTangle:

- ✅ Suggests improvements → You choose whether to accept
- ✅ Flags potential issues → You decide whether to address them
- ✅ Offers alternatives → You select what works best
- ❌ Does NOT automatically modify your requirements
- ❌ Does NOT make decisions on your behalf
- ❌ Does NOT operate without your explicit action

​

6. AI Limitations & Risks

 

When NOT to Rely on AI

Do not use ReqTangle's AI features for:

🚫 Safety-critical systems - Medical devices, aviation, automotive safety  
🚫 Legal compliance - Regulatory requirements, contractual obligations  
🚫 Security-critical specifications - Cryptography, authentication, access control  
🚫 Final decision-making - AI assists; humans decide  

​

Potential Errors & Biases

AI suggestions may:

- Introduce subtle inaccuracies or ambiguities
- Reflect cultural or linguistic biases from training data
- Misinterpret specialized industry or domain terminology
- Suggest changes that alter intended meaning
- Fail to understand your specific project context

​

Best Practice: Treat AI suggestions as a "second opinion" from a junior colleague, helpful but requiring verification.

​

7. EU AI Act Compliance

 

Regulatory Classification

Under the EU Artificial Intelligence Act (Regulation 2024/1689):

- ReqTangle's AI System: General-purpose productivity tool
- Risk Level: Not high-risk under Article 6(2)
- AI Provider: Codicent Inside AB (Article 3, definition 3)
- AI Deployer: Ikizler AB / ReqTangle (Article 3, definition 4)

 

Transparency Obligations (Article 52)

You are hereby informed that:

1. You are interacting with an AI system when using ReqTangle's AI features
2. Content generated or influenced by AI is clearly labeled in the user interface
3. AI outputs are not authoritative and require human review
4. You can access this transparency information at any time

 

Technical Documentation (Article 11 & 53)

ReqTangle maintains comprehensive technical documentation about our AI system, including:

- System design and development process
- Risk assessment and mitigation measures
- Data governance and quality metrics
- Human oversight mechanisms
- Performance testing and accuracy evaluations

This documentation is available to EU competent authorities upon request.

​

Rights & Safeguards

Under the EU AI Act, you have the right to:

- Understand AI decisions - Know how suggestions are generated
- Human review - Challenge or question AI outputs
- Not be subject to solely automated decision-making - Humans always in the loop

​

8. Security & Privacy

​

Data Protection

All AI processing complies with GDPR:

- Legal Basis: Contract performance (Article 6(1)(b)) and your explicit use of features
- Data Minimization: Only necessary requirement text is processed
- Purpose Limitation: Data used solely to provide AI suggestions
- Storage Limitation: Processing data not retained beyond session

For details, see our [Privacy Policy](https://www.reqtangle.ai/privacy-policy).

 

Security Measures

AI processing infrastructure includes:

- Encryption in transit (TLS 1.3) and at rest (AES-256)
- Access controls and authentication
- Regular security audits and penetration testing
- Incident monitoring and response

 

9. Feedback & Improvement

 

How We Improve AI Quality

We improve ReqTangle's AI features through:

✓ Aggregate usage analytics - Anonymized patterns of feature usage  
✓ User feedback - Ratings and comments you provide on AI suggestions  
✓ A/B testing - Comparing different AI approaches (with your consent)  
✓ Manual review - Engineering team reviews flagged AI errors  

​

We do NOT:

❌ Review individual user requirements without explicit consent  
❌ Use your data to train models for other users  
❌ Share your usage patterns with third parties  

​

10. Reporting AI Issues

 

Types of Issues to Report

Please report:

- Inaccurate suggestions - AI recommendations that are clearly wrong
- Harmful outputs - Offensive, biased, or inappropriate content
- Privacy concerns - AI accidentally revealing information it shouldn't know
- Security issues - Vulnerabilities related to AI features

​​

How to Report

For quality issues or feedback: 
Email: info@reqtangle.ai

 

For privacy concerns:  
Email: dpo@reqtangle.ai

 

We take all reports seriously and will:

- Acknowledge receipt within 48 hours
- Investigate and respond within 5 business days
- Implement fixes or safeguards as needed
- Notify you of the outcome

 

11. Future Developments

​​

Commitment to Transparency

As we add new AI capabilities:

- ✅ We will update this policy and notify you
- ✅ New features will be opt-in by default
- ✅ We will explain how they work and what data they use
- ✅ You will always control your data and AI usage

​

12. Contact & Questions

​

AI Transparency Inquiries

For questions about ReqTangle's AI features:

​

General Questions:
Email: info@reqtangle.ai  
Website: https://www.reqtangle.ai/ai-transparency

Data Protection Officer:
Email: dpo@reqtangle.ai  
Postal: Ikizler AB, Hageby Allé 21 b, 141 69 Huddinge, Sweden

Platform Provider (AI Technology):
Codicent Inside AB  
Email: dpo@codicent.ai

​

Regulatory Information

GDPR Supervisory Authority (Sweden): 
Integritetsskyddsmyndigheten (IMY)  
Email: imy@imy.se  
Website: https://www.imy.se

​

14. Version History

| Version | Date         | Changes                             |
| ------- | ------------ | ----------------------------------- |
| 1.0     | Feb 11, 2026 | Initial publication for beta launch |

​

Next Review: May 11, 2026 (or upon material changes)

​

Acknowledgment

By using ReqTangle's AI features, you acknowledge that:

- You have read and understood this AI Transparency Policy
- You understand AI suggestions are assistive, not authoritative
- You will apply human judgment and verify all AI outputs
- You accept the limitations and risks described in this policy

​

Thank you for helping us build responsible AI for requirements management.