AI Transparency
How Deventura uses AI to analyze software delivery signals, classify AI-assisted work, and generate analytical insights.
Last Updated: May 13, 2026
Intended Use
Deventura uses AI to support analysis of engineering activity data, software delivery signals, and AI-assisted work. Deventura's AI-generated insights are not intended for use in employment, promotion, compensation, or disciplinary decisions. Any deployment for such purposes may require high-risk compliance measures under the EU AI Act.
How We Use AI
Deventura uses AI to analyze engineering activity data and software delivery signals, and to generate analytical insights about AI-assisted work, delivery outcomes, and patterns over time. Our AI analysis is generated using Claude by Anthropic, a large language model designed with safety and reliability in mind.
Activity Analysis
AI analyzes merge request data across multiple dimensions to surface software delivery signals based on code complexity, delivery consistency, and collaboration patterns.
AI-Assisted Work Detection
Deventura classifies each analyzed contribution as Near-certain AI, Likely AI, Ambiguous, Likely human, or Near-certain human, with a separate low, medium, or high confidence level. Contributions are skipped when the signal is not clear enough.
Trend Detection
AI identifies patterns across time periods to help analyze how AI initiatives, agents, tool changes, or process changes affect delivery outcomes.
Human Oversight & Decision-Making
All final decisions are made by humans. Deventura's AI provides recommendations and insights, but does not make autonomous decisions about hiring, promotion, compensation, or termination.
Managers review and interpret all AI recommendations before acting. All insights are advisory and reviewed by humans — AI does not make decisions on its own.
- Managers review and interpret all AI recommendations before acting on them
- AI scores are inputs, not verdicts — they inform human judgment
- Developers can provide context that AI cannot understand
- Human managers make all people decisions using AI insights as one data point among many
- Final decisions remain human — AI supports, managers decide
Data Privacy & Security
We take data protection seriously and comply with GDPR, CCPA, and other privacy regulations.
- No code is stored — only metadata from merge requests is analyzed
- All data is encrypted at rest and in transit
- You control your data — delete it anytime
- Read-only access to repositories — we never write or modify code
- Customer data processed by Deventura is not used to train AI models
For more details, see our Security & Compliance page and Data Processing Agreement.
AI Limitations & Transparency
We acknowledge the following limitations of our AI system:
- Context limitations: AI cannot understand organizational politics, personal circumstances, or non-technical factors
- Metric-based: AI analysis is based on quantifiable signals from engineering activity data and may not capture qualitative contributions or factors outside the data
- Historical bias: AI learns from past patterns, which may contain historical biases that we actively work to identify and mitigate
- Not a replacement for management: AI augments human judgment but cannot replace the nuanced decision-making of experienced leaders
Fairness & Bias Mitigation
We take active steps to ensure our AI system is fair and unbiased:
- Objective metrics: We measure code and contributions, not personal characteristics
- Regular audits: We continuously monitor for bias in our algorithms
- Diverse training: Our AI is trained on diverse codebases and team structures
- Transparent methodology: How AI-generated insights are calculated is documented and reviewable
- Appeal process: Any AI-generated insight can be questioned and reviewed
EU AI Act Compliance
Deventura is designed to comply with the EU Artificial Intelligence Act. Our system is classified as a limited-risk AI system under the Act, and we fulfill all transparency obligations.
- Transparency: Users are informed when interacting with AI
- Human oversight: All high-stakes decisions require human review
- Documentation: We maintain comprehensive technical documentation
- Risk management: We conduct ongoing risk assessments
Your Rights
As a user of Deventura, you have the right to:
- Understand how AI-generated insights about you were produced
- Contest any AI-generated insight or recommendation
- Request human review of any AI decision
- Opt out of AI analysis (though this limits product functionality)
- Access your data and request deletion at any time
Contact Us
If you have questions about how we use AI, want to exercise your rights, or need to report a concern, please contact us:
privacy@deventura.com — Subject: "AI Transparency Inquiry"
Want to understand how Deventura uses AI?
Book a demo and we'll explain how Deventura uses AI to analyze engineering activity data, classify AI-assisted work, and generate analytical insights.
Book a demo