In today’s development environment, enabling AI agents to write, review, and deploy code must not come at the expense of your IP or other security considerations. At CurieTech AI, we built our platform from the ground up to be secure, transparent, and enterprise-ready—so you can use powerful coding agents with confidence.
Our Core Security and Trust Pillars
At CurieTech AI, security isn’t an afterthought, it’s the foundation. Every layer of our platform, from model interaction to code execution, is designed to protect your intellectual property while maintaining full transparency and control. We combine enterprise-grade governance with cutting-edge AI capabilities, giving organizations the confidence to innovate faster without compromising data privacy, compliance, or trust.
Code Security
- No Data used for Model Training: Your code and data remain entirely your property, CurieTech AI does not use your proprietary repositories or artifacts for model training or data enrichment.
- Full Control over your Data: Your code is used only during the duration of the task with the ability to configure a code cache time-to-live (TTL) policy. This automatically removes any cached copy of your codebase once your task is complete or after the defined TTL expires, ensuring you are in the driver’s seat.
- Trusted Integrations: CurieTech AI provides trusted integrations with leading version control systems, all connected through OAuth-based authorization ensuring safe and auditable access.
Zero Data Retention
- ZDR Agreement with Model Providers: We have a ZDR agreement with all AI model providers ensuring any data sent to them is never stored or used for training purposes.
- Control over Context sent to Models: CurieTech’s advanced retrieval engine minimizes the amount of code context, including secure credentials, to be sent to the model providers that is required for the task to be completed.
- User-Controlled Data Deletion: We provide an ability to delete your tasks and conversations in CurieTech to ensure none of your prompts or Curie generated assets are persisted.
Built-In Guardrails for Code Generation
- Dynamic Grounding: Curie’s advanced reasoning agents are grounded in MuleSoft’s domain ensuring the code produced has a high degree of accuracy and do not introduce unintended bugs, follow the best coding practices for your organization which increases trust and makes it easier to maintain the generated software.
- Validity Check and Safety Filters: All code generated by Curie is validated against a set of strict guidelines for correctness, completeness, efficiency, and coding best practices following your enterprise standards.
- Strict Code Reviews: We embed code review agents in CI/CD workflows to evaluate AI-generated pull requests for security, style, or policy violations.
- Prompt Guardrails and Defense: We have strong guardrails against prompt injection attacks to avoid the agent from performing unintended actions, executing bad terminal commands, etc.
Secure Code Execution
- Private Runners: You can deploy private runners which are sandboxed runtimes that run inside your own network or cloud environment under your governance which the agents can use to test their work (including custom connectors/jars).
Encryption & Infrastructure Hardening
- Data Encryption: All data is encrypted in transit and at rest using industry best practices such as AES and TLS protocols to protect your data at every stage.
- Audit Trail: Our production systems are continuously monitored through logging and alerts are set in place for unusual application states.
Identity, Access & Governance
- Single Sign-On (SSO) and Role-Based Access Control (RBAC) allow precise control over who can connect repos, run agents, or configure settings.
- Multi-Factor Authentication (MFA) is enforced to protect user accounts from unauthorized access.
- Admins can set workspace-level policies, session timeouts, role scopes, and version controls.
In addition to the above, CurieTech is SOC 2 Type II certified which includes a third-party audit verifying the effectiveness of our security, availability, and confidentiality controls.
The CurieTech AI Promise
With CurieTech AI:
- Your code and data stays private.
- Code Execution happens where you control it.
- Access, roles, and policies are clear, auditable, and enforced.
- AI-generated code is treated as first-class software that must pass security checks.
.jpg)
