Bezi AI
The bet is that engine context beats model quality for game work, and it holds up: an assistant that can read your scene graph, run in play mode and check its own edits is doing something a general code assistant cannot. Credit metering is the friction - an agent that iterates against a running editor burns through an allowance faster than autocomplete does, and the tiers escalate steeply.
Pros
- Runs inside the Unity Editor with scene indexing and play-mode debugging, not as an external chat
- Context spans design docs, spreadsheets and git history alongside code
- Reusable Skills capture repeated workflows instead of re-prompting each time
- A memory layer keeps project conventions across sessions
Cons
- Unity only - nothing here transfers to another engine
- Credit metering makes cost unpredictable for agent-mode work that iterates
- The jump from Pro to Advanced triples the price for three times the credits, with no middle step
- No company registration, country or founding year is published