Security and Data Governance

• Built with Test Data

  AI writes and tests the code with synthetic or redacted data. Your records are never sent to an AI model during the build.

• The Code Does the Work

  Our deliverables read, sort and draft the same way every time, inside your systems.

• AI in Live Systems

  Where a running system uses AI, it goes through Anthropic’s business API, under terms that do not use the data to train AI models. Consumer chat applications are never used.

• Sensitive Fields Excluded

  Details such as Social Security numbers and bank accounts are masked or excluded from what the AI sees.

• Human Approval

  A person approves anything that moves money, changes a record or commits you to a customer.

• A Complete Activity Record

  Every action the system prepares is logged with who approved it and when.

• Your Systems, Your Data

  Wherever possible we build inside your own environment, such as Microsoft Azure, and keep no copies.

• Least-Privilege Access

  Separate accounts with only the access the work requires, removed when the engagement ends.

• Read-Only During Design

  Nothing changes in your systems during Solution Design.

• Secrets in a Vault

  API keys and passwords stay server-side, for example in Azure Key Vault, never in code.

• NDA on Request

  We sign a non-disclosure agreement or a data processing agreement on request.

• Returned or Deleted

  At the end of an engagement, your data is returned or deleted.

Please send them to alexp@khipuai.co or raise them on a free discovery call. Please do not send passwords or client records by email; we will first agree on a secure way to share them.

We are documenting these practices as formal policies and procedures for your IT team to review; each will be listed here once complete. How this website handles your information is described in the privacy policy.