Security and privacy at Usedge
Usedge protects the research data its customers entrust to it with a security program aligned with ISO 27001 and SOC 2, hosted entirely in the European Union.
Last updated
AWS Paris
eu-west-3 · three availability zones
ISO 27001 & SOC 2
Security program aligned with both
GDPR
Processor for our customers · DPA on request
Encrypted
At rest with AWS KMS · TLS 1.2+ in transit
Our security program follows four principles
Least privilege
Access is limited to people with a legitimate business need.
Defense in depth
Security controls are layered, so no single failure exposes data.
Consistency
Controls apply across the whole company, product and infrastructure.
Continuous improvement
Controls are monitored continuously and reviewed by management at least once a year.
Infrastructure and hosting
The Usedge platform runs on Amazon Web Services in the Paris region (eu-west-3), across three availability zones.
Containers
The application runs on Amazon ECS with AWS Fargate, inside a private network (VPC) with security groups that open only the ports each service needs.
Edge protection
All traffic passes through Cloudflare, and the load balancers accept requests only from Cloudflare.
Encryption at rest
All databases and storage buckets holding customer data are encrypted. Encryption keys are managed in AWS Key Management Service, which keeps key material in hardware security modules.
Encryption in transit
Data travelling over public networks is protected with TLS 1.2 or higher. Certificates are managed by Cloudflare and AWS.
Secrets
Application secrets are encrypted and never stored in source code.
Backups
Customer databases are backed up automatically and the backups are encrypted.
Physical security of the data centers is the responsibility of AWS, which holds ISO 27001 and SOC 2 reports that Usedge reviews every year.
Access and authentication
For customers
Single sign-on
Customers can sign in through Okta or Microsoft. A workspace can require single sign-on for all its members.
Brute-force protection
Sign-in, password reset and API endpoints are rate limited.
Roles and permissions
Workspace administrators control who can view or edit each team and project.
For Usedge staff
Google Workspace sign-in
Staff sign in through Google Workspace.
Restricted production access
Access to production is limited to the engineering team, granted by role, and reviewed regularly.
Offboarding
Access is removed when someone leaves the company.
AI and Usedge Chat
Usedge Chat lets users analyze their research data with the large language model of their choice. It is built on LibreChat, an open-source framework, and runs in Usedge's own AWS environment.
| Model provider | Role | Processing location |
|---|---|---|
| OpenAI | Subprocessor | European Union |
| Anthropic | Subprocessor | European Union |
| Mistral AI | Subprocessor | European Union |
Only what the question needs
A model receives the prompt and the data a user selects, not the whole workspace.
No training on customer data
We use the providers' business APIs, whose terms do not allow customer data to be used to train their models.
Permissions apply
Usedge Chat only reaches data the user is already allowed to see.
Bring your own model
Customers can connect their own model provider instead. In that case, the customer's agreement with that provider governs the data sent to it.
Data protection and privacy
GDPR
Usedge is a French company and processes personal data under the GDPR, as a processor for its customers. A data processing agreement is available on request.
Data location
Customer data is stored in the European Union (AWS Paris).
Retention and deletion
Customer data is kept for the duration of the contract and deleted on request or when the contract ends, following our data retention schedule.
Subprocessors
We review the security of every subprocessor before use and every year after. Our main subprocessors are AWS (hosting), Cloudflare (network protection) and the AI model providers listed above. Customers are informed before a new subprocessor is added.
Security program
Continuous monitoring
Our controls are monitored continuously with Vanta. AWS GuardDuty detects intrusions, and our infrastructure and application are monitored with CloudWatch, Grafana and Sentry.
Penetration testing
An independent firm tests the platform every year. All findings from the August 2025 test have been fixed; the 2026 test includes Usedge Chat.
Vulnerability management
Our cloud environment and container images are scanned for known vulnerabilities, and fixes are prioritized by severity.
Policies
Our information security policies are reviewed and approved by management every year, and every employee and contractor must accept them.
People
Staff and contractors sign confidentiality agreements and must complete security awareness training.
Incident response
A documented incident response plan defines how we detect, contain and report security incidents, including notifying affected customers.
Contact
To report a vulnerability, ask a security question or request our data processing agreement, write to [email protected]. We acknowledge vulnerability reports within 2 business days.