AGKit vs Firebase AI Logic
Choose AGKit when you want application tools and an agent runtime backed by Cloud-managed profiles, model routes, credentials, and budgets.
Firebase supplies mobile and web SDKs for Gemini capabilities, with App Check, per-user rate limits, and remote configuration.
 AGKit | Firebase AI Logic | |
|---|---|---|
| Application stack | Swift, Android, React Native, and Web; one Cloud backbone | Apple, Android, web, Flutter, and Unity SDKs |
| Application actions | Agent runtime, tool contracts, permission hooks, and cancellation | Function calling and multimodal Gemini capabilities |
| Configuration | Cloud-managed model tiers, credentials, and agent profiles | Remote Config and server prompt templates (Preview) |
| Production controls | End-user identity, project/user budgets, and configured attestation | App Check, per-user rate limits, authentication options |
| Application responsibility | Tool implementation, state validation, approvals, and actual undo | App implements its actions and state lifecycle |
Choose your platform and access path for SDK availability and setup. Guard and confirmation support depends on the SDK execution path you integrate.
Where Firebase fits
You want Gemini in a Firebase application. App Check, authentication options, and rate limits support the client integration; Remote Config and server prompt templates let you update configuration remotely.
Where AGKit fits
You want a runtime for application tools together with Cloud-owned model tiers, profiles, credentials, and budgets. Your app names its intent while Cloud resolves the configured provider and model.
What your team implements
Your application defines its actions and the state they can change. Build approval and undo around those operations, then test cancellation and concurrent edits against your real data model.