Meet Emma, your AI analyst
Use case · Private AI and cost

Run AI privately, at a cost you control

Give every team private AI that never trains on your data, put a budget on every key, and see where your cloud money goes.

What you get

Private by default, no surprises on the bill

AI your security team can say yes to, and spend your finance team can follow.

Private models

Models run inside secure enclaves. Prompts and answers stay private and are never used for training.

Keys with limits

Each team or app gets its own key, with a monthly budget and a per-minute limit.

Spend in view

Cloud and AI costs by team, service and tag, with alerts on unusual jumps.

With Emma

Ask where the money went.

Emma reads your cloud and AI bills, explains the changes and suggests savings. Changes still wait for your yes.

Explore Emma →

Private AI for every team

Point the OpenAI or Anthropic SDK at your datagab endpoint. Your code stays the same, your data stays private.

  • Works with the OpenAI and Anthropic SDKs
  • Hardware-protected memory you can verify
  • Confidential servers with SSH for your own workloads

A budget on every key

Stop surprises before the bill arrives. Every request is metered, and keys stop at the limit you set.

  • Monthly budgets per key and team
  • Per-minute limits
  • Cost per agent, key and model

Cloud savings with one approved click

See spend by team and service, get told about unusual jumps, and act on savings once a person approves.

  • Budgets and anomaly alerts
  • Stop idle and resize oversized resources
  • Shared costs split by tag and team

Private AI, with the bill in view.

See private AI and cost control on your own accounts.

FAQ

Frequently asked questions

Is our data used to train models?

No. Models run inside secure enclaves, and prompts and answers are never used for training.

Do we need to change our code?

No. Point the OpenAI or Anthropic SDK at your datagab endpoint and use a datagab key.

Which clouds are covered for cost?

AWS, Azure and Google Cloud, with costs split by team, service and tag.