The feature described in this article is currently in closed beta and available to a limited group of users. Some details may change before it becomes available to everyone.
Auto-Labels automatically organize your incoming emails into categories based on what each message requires from you: whether it needs a reply, is informational only, or has already been actioned. This helps you see at a glance which emails to handle first and which can wait, without sorting through your inbox manually.
What Auto-Labels do
Auto-Labels build on top of Spark's Smart Inbox. Smart Inbox already separates real emails from newsletters and notifications. Auto-Labels go a layer deeper — giving you finer granularity within those real emails so you know exactly what each one needs from you.
There are seven labels, and each one answers a specific question about a message:
- To respond — emails you need to reply to
- FYI — emails that don't require a response but may be important
- Comment — comments and mentions from Google Docs, Microsoft Office, Figma, and other collaborative tools
- Marketing — marketing messages and promotional emails
- Meeting update — calendar notifications from Zoom, Google Meet, and similar services
- Notification — automated updates from the tools you use
- Actioned — threads that have been resolved
Prerequisites
In order to use Auto-Labels, make sure that all of these conditions are met:
- Spark Desktop is running on your device: right now, the feature is available on Mac and Windows devices
- Spark Pro subscription is active: this feature is a part of the Pro plan
- Spark +AI is enabled in Spark Settings
- Auto-Labels toggle is on in Spark Settings > Spark +AI > Inbox Automation
- Personal account is used, and it’s Gmail: the feature won’t work for shared accounts and works for Gmail accounts only
How to enable Auto-Labels
You will have a chance to enable this feature during the What’s New onboarding process. If you skip it, follow these steps to start using the feature on your Mac or Windows:
- Launch Spark Desktop on your device.
- Go to Settings > Spark +AI > Inbox Automation.
- Enable the Auto-Labels toggle.
- Choose the email accounts you want to use the feature with and toggle them on.
- Click the arrow button next to the email account to see the list of labels.
- Select the labels you need and toggle them on.
After this, Spark will automatically apply labels to all the emails you’ve received for the past 7 days.
How it works
Once you enable Auto-Labels, Spark gets to work on two fronts:
- Every incoming email is automatically labeled as it arrives.
- Spark labels your recent emails, too — up to 300 messages from the last 7 days.
Only emails that match a label's definition will get tagged: Spark won't force a label onto a message that doesn't fit. Also, Auto-Labels look and behave exactly like the labels you'd apply manually, so you won’t be confused. Each email receives one Auto-Label to keep your inbox clean, so your emails won’t have more than one auto-label applied to them.
You’re still in charge of Auto-Labels, so you can:
- Remove a label from any email if it doesn't fit
- Change a label's color to suit your workflow
- Send feedback when something looks wrong, if an auto-label was applied incorrectly, or an email should have been labeled but wasn't
- Disable specific Auto-Labels you don't want to use
- Turn the feature off entirely for a specific account, or across the board
- Choose and set up different labels for different accounts, if needed
You can disable the feature entirely under Spark +AI Settings > Inbox Automation. You can also disable Auto-Labels for a separate email account there.
Correcting Auto-Labels
If Spark applies the wrong label, you can remove it and apply the correct one manually. Spark doesn't automatically learn from your corrections, but you can help our team improve accuracy by reporting the issue through the feedback pop-up.
How to report a wrong label
There are two ways to send feedback:
- Through the Command Center
- Open the affected email thread.
- Press ⌘K (Mac) or Ctrl+K (Windows).
- Choose Give feedback on Auto-Label.
- Pick the label Spark should have used, and optionally add a comment. Your feedback goes directly to our AI team.
- From the label itself
Right-click the label and select Wrong label?
If you manually add a label that matches an Auto-Label name to an email that didn't receive one, Spark will suggest you share feedback so our team can investigate why it was missed.
Does Spark learn automatically from the way I correct labels?
If you remove a label or change it, that change stays on your email in Gmail — it does not silently retrain a model or personalize Auto-Labels behind the scenes. The only time your corrections are seen by our dedicated team is if you explicitly submit them through Command Center > Give feedback on Auto-Label and share the email explicitly with us (mark the checkbox “Share this conversation with Spark to improve Auto-Labels”).
For more details, please review our Privacy Policy, which explains how your data is handled and protected.