Transparency in AI: How Remesh Builds Trust Through Responsible Implementation

As artificial intelligence becomes increasingly integrated into research platforms, organizations need clarity on how these technologies work, what data they use, and how they're governed. At Remesh, we believe transparency isn't just good practice, it's essential for building trust with our customers and participants.

Two Types of AI but One Commitment to Control

Remesh's AI capabilities fall into two distinct categories with each designed for specific use cases and safeguards in mind.

Our proprietary machine learning models are developed entirely in-house and trained specifically for your data without external pre-training bias and used for the sole benefit of each individual client. These advanced models analyze participant responses and engagement patterns to identify consensus and highlight representative insights. All models and their outputs remain specific to individual conversations within single workspaces and are never shared across clients.

Our Generative AI features are optional and require explicit customer opt-in. These capabilities, powered by industry-leading third-party providers, enhance the analysis and usability of research conversations through features like response summarization, discussion guide improvements, auto-coding of open-ended responses, and simulated participants for practice sessions.

Privacy and Security by Design

Data protection forms the foundation of our AI implementation. We don't intentionally send personally identifiable information (PII) to GenAI systems, and participant response data is pseudonymized and de-identified by default. However, we maintain transparency about limitations—participants may voluntarily include personal details in their responses, and this information isn't automatically scrubbed prior to processing. 

All data transmission and storage use enterprise-grade encryption, and data is processed and stored exclusively in the United States. Our third-party providers retain data for limited periods for security and abuse monitoring, after which it's automatically deleted. Crucially, these providers don't use customer data to train or improve their models.

Human-Centered AI Governance

Our approach to AI governance centers on human oversight and control. AI-generated output requires human review or "human-in-the-loop" functionality. GenAI features are clearly marked in the platform interface, ensuring users know when AI is involved.

A dedicated ethics committee comprising representatives from multiple departments reviews use cases and policies. Our development practices align with industry security frameworks, including comprehensive protections against common AI vulnerabilities. Remesh’s Chief Technology Officer and Head of Security and Compliance meet regularly to ensure best practices are followed.

Risk Mitigation Through Technical Safeguards

We've implemented multiple layers of protection to ensure AI output accuracy and reliability. Advanced optimization techniques and structured selection processes help ensure outputs remain accurate and aligned with source material. Features that generate text outputs are designed to minimize inaccuracies and stay true to the original data.

These features undergo regular review for bias and errors. Processes incorporating AI are regularly evaluated for risk to users, data quality, and output accuracy.

Customer Control and Ownership

Customers retain complete ownership of their data and AI outputs. The Remesh platform includes self-deletion features that allow customers to remove their conversation data and models at any time, ensuring compliance with applicable laws and internal policies.

The platform functions fully without GenAI features. When customers do choose to enable these features, they maintain direct control over acceptance and usage of all AI-generated outputs.

Continuous Evolution and Transparency

We regularly update our underlying AI models based on internal evaluation and testing to mitigate bias, identify anomalies, and ensure alignment with customer expectations. Our technology infrastructure is built using modern development frameworks and follows industry best practices, including SOC 2 Type II. We maintain updated documentation of our service providers and their specific functions for complete transparency.

Building Trust Through Documentation

Understanding how AI works in our research platform shouldn't require guesswork. We've created comprehensive documentation covering AI implementation, governance practices, and technical safeguards. This documentation includes detailed FAQs, security specifications, and guidance for configuring features within your organization.

For organizations evaluating AI-powered research platforms, the key questions aren't just about capabilities, they're about control, transparency, and trust. Our approach prioritizes customer control over AI features, transparency in implementation, and rigorous governance practices that protect both data and participant privacy.

For detailed technical specifications, governance practices, and implementation guidelines, contact hello@remesh.ai

About the Author

With 18+ years in software and technology, Ross Coudeyras is a recognized authority in cybersecurity and data privacy. Currently heading Security and Compliance, as well as serving as Data Protection Officer at Remesh, he is known for his practical approach to securing digital landscapes. At Remesh, Ross leads security and compliance efforts, ensuring the organization aligns with regulations while staying resilient against cyber threats. His real-world insights bridge the gap between security measures and compliance needs.  As an adjunct instructor at Doane University, Ross imparts hands-on knowledge to future cybersecurity professionals, leveraging his industry know-how to prepare students for real challenges.

‍