Gmail Connector
Gmail communication data and categorized email signals.
Updated 19 March 2026
gmailconnectoremailmobile
Gmail Connector
Gmail communication data and categorized email signals.
Connector Metadata
| SDK connector id | gmail |
| Works on | React Native, Swift, Mobile OAuth |
| Current collection surface | Mobile OAuth + collection |
| Information density | Very high |
| Signal quality | Very high |
| Best for | Communication style, life context, transactional preferences, recurring interests, and high-density intent signals. |
Activation
To show this connector in the SDK, add gmail to the platform allowlist for the SDK surface you are using.
Web
This connector is not currently documented for the web connector surface. Use one of the supported mobile surfaces below.
React Native
<OnairosButton
AppName="YourApp"
allowedPlatforms={['gmail']}
recommendedPlatforms={['gmail']}
onComplete={(result) => console.log(result)}
/>
Swift
let config = OnairosConfig(
apiKey: "your_api_key",
platforms: [.gmail],
recommendedPlatforms: [.gmail],
urlScheme: "yourapp",
appName: "YourApp"
)
Connector ids are lowercase
Use gmail in allowedPlatforms. Display names such as Gmail may work on some SDK surfaces, but lowercase connector ids are the safest cross-platform choice.
Current SDK Surface
Mobile OAuth + collection
What We Collect
- Categorized emails using Gmail's built-in categories plus subject heuristics for likely likes vs dislikes.
- Email headers, snippets, and message body text for categorized email documents.
- Sent-email content after stripping quoted replies/signatures and scrubbing PII before training.
Current Onairos Handling
- Collected through the Gmail connection flow.
- Additional categorization and cleanup steps prepare inbox and sent-email data for training.
Notes
- The categorized-email route explicitly groups primary/updates-style mail as likely positive signals and spam/promotions-style mail as likely negative signals.
- The sent-email training path removes quoted replies and auto-generated content before sentiment analysis.