CaseMark hands firms the keys to their own AI models
You can now wire Mike straight into your own Anthropic or Gemini account instead of routing everything through CaseMark's gateway.
Until now, anyone running this fork of Mike reached the AI models through CaseMark's shared gateway. This change lets a firm plug in its own credentials with Anthropic, the company behind Claude, and Google's Gemini - so the bill, the contract, and the data relationship sit directly between you and the model maker rather than through a middle layer.
The plumbing is built for people who worry about credentials. Keys are encrypted where they're stored, and CaseMark checks each one actually works before trusting it. Deleting a key doesn't wipe the record - it marks the key revoked and keeps the audit trail. And in production, the old habit of quietly falling back to a shared house key is switched off, so a firm only ever uses the credentials it deliberately set up. It's all managed from a new models page in account settings.
Spotted something wrong? Or know the PR text has fresher detail than the writeup above?