Pricing
Per person, per month.
Not a percentage of your Bedrock spend. A product whose job is to reduce that spend should not be paid more when it grows.
You pay AWS for Bedrock directly, as you do today. TokenLock is licensed per person and runs on your own infrastructure, so there is no margin on inference and no data leaving your account.
Billed on people with spend in the month. Someone who did not use Claude that month is not a seat.
- Attribution by person, team, project and application
- Budgets that block or fall back, hourly to monthly
- Application keys with per-end-user budgets
- Extractions — classify conversations into your own dimensions
- Entra ID and Google Workspace federation
- Managed-settings rollout for Claude Code and Claude Desktop
- Deploy guide, header reference and rollout support
Private beta. Pricing is fixed for the first year of any beta agreement.
Worth a call if
You want Claude for your staff, and your security team needs it to run inside your AWS account.
Your Claude spend runs through Amazon Bedrock and appears as one line on the AWS bill.
Somebody has asked who spent it, or what it was for, and the answer was not available.
Claude Code is out to a team or a department, and growing.
A security review already approved Bedrock, and you would rather not reopen it.
Not us, and we will say so
You need OpenAI, Gemini and Bedrock behind one endpoint. That is LiteLLM.
Your spend is on the first-party Anthropic API rather than Bedrock. The Console covers more of your case than we do today.
You want per-person productivity scoring. We are a spend tool and will not build that.
If you need OpenAI, Gemini and Bedrock behind one endpoint, use LiteLLM — genuinely, it is good at that. TokenLock is for organisations whose AI spend is Bedrock and whose problem is explaining it.
See it on your own spend.
A pilot rather than a demo. Point one team's Claude Code at TokenLock and the first breakdown is there within the hour.