Every AI coding rollout reaches the security review. These are the questions that stall it.
Which account do prompts run through, and who sees the invoices for that spend?
If we move to another model provider, does our tooling, history and cost reporting come with us?
Are they encrypted at rest with AES-256-GCM, how are they rotated, and which roles can use them?
Paste your Anthropic API key once in the admin settings. It is encrypted with AES-256-GCM before it ever touches disk, and every request runs through your own account, so the spend lands on your Anthropic billing, not ours.
Bring your own key for Anthropic, OpenAI, Google Gemini or AWS Bedrock. Token costs from every provider are normalized into one spend view, so finance sees one number and engineering sees the breakdown by model.
Connect GitHub in one click for commit and pull request data. Sign in with Google Workspace SSO restricted to your domain, or Okta and Azure AD on Enterprise, with four roles enforced on the server.
Terraform deploys the full CloudByte AI stack into your AWS account, isolated in your VPC. Use Bedrock for models, your own KMS for keys, and run fully air-gapped if you need to.
What BYOK changes for engineering leadership, and what it gives your auditors.
For CTOs and VPs of Engineering
For Security and Compliance
Bring your own key means your organization plugs in its own Anthropic API key. Every call bills to your Anthropic account, and CloudByte AI never proxies billing. The key is encrypted at rest and only decrypted for the request that needs it.
CloudByte AI normalizes token usage from Anthropic, OpenAI, Google Gemini and AWS Bedrock into a common cost-per-request metric, so spend from every provider adds up in one view and can be broken down by model, developer and project.
Keys are encrypted with AES-256-GCM and stored in your organization's Postgres row with a unique nonce per row. The encryption key lives in AWS Secrets Manager on the hosted service, or in your own KMS when you self-host. Keys are never logged.
Yes. Prompts reach Anthropic through your own account, inside your data boundary. Every key use, rotation and failed authentication is written to an audit log you can export for your auditors.
Yes, on the Enterprise plan. The stack is containerized and deployed with Terraform into your AWS account, and it can run air-gapped with AWS Bedrock inside your VPC.
Plug in your own Anthropic key and see every session, spend line and audit event in your own workspace.
