One login, three tools, weekly credits already loaded. Built on open models you're not locked into — the same stack the AI in Practice course teaches on.
Credits refill every week. Nothing you type is used to train anyone's model. Cancel anytime.
Taking a course? It's included. Enter the email you enrolled with and we'll send your sign-in link — no card needed.
Every piece of the workspace is open source and self-hosted. You could run the same stack yourself — the course shows you how. What you pay for is the curation, the credits and the setup you don't have to do.
A curated set of open frontier models — Kimi K2, DeepSeek, Qwen and others — routed through our own gateway. "Auto" picks the best one for your prompt.
Roughly $2 of model spend a week, refilled automatically. That's hundreds of long conversations; if you hit the ceiling before the refill, the workspace tells you.
Conversations live in our own database, not a vendor's. Model providers see requests but no account. You can turn memory off per account.
The course app and the AI workspace are different systems. Same email, two logins. Your membership unlocks both; cancelling the course ends both.