Try experimental Gmail features

Experimental features are pre-releases that you can turn on to test and share feedback about. You can opt in to experimental features in your Gmail settings. 

Get access to experimental features

You can only turn on access to experimental features from your computer using the new Gmail.

Note: Google does not offer any support for experimental features. Google reserves the right to change these services or even remove them at any time. 

To turn on Experimental Access, you'll need to use your computer. 

  1. In the top right corner, click SettingsSettings and then Settings. If you haven’t started using the new Gmail yet, click Try the new Gmail
  2. Under “General,” scroll down to Experimental Access.
  3. Click the box to Enable experimental access.
  4. At the bottom of the page, click Save changes.

You now have access to experimental features.

Change settings for experimental features

When you have access to experimental features, you’ll be able to customize the settings for each from the Settings. To change settings for experimental features:

  1. In the top right corner, click SettingsSettings and then Settings. If you haven’t started using the new Gmail yet, click Try the new Gmail
  2. Under “General,” scroll down to look for the features that have the experiment iconExperiment.
  3. Next to each feature, click choose to turn it on or off.

Available experiments 

See the following list for experiments that are currently available.  

Use Smart Compose

You can let Gmail help you write better emails, faster. The Smart Compose feature is powered by machine learning and will offer suggestions as you type. To accept a suggestion, press Tab.  

To turn Smart Compose on or off:

  1. From the Gmail, click SettingsSettings and then Settings
  2. Under “General”, scroll down to Smart Compose.
  3. To turn on predictions, click to choose Writing suggestions on.  To turn off predictions, click to choose Writing suggestions off.  

Note: Smart Compose is available in English. Smart Compose is not designed to provide answers and may not always predict factually correct information.

About machine learning

As language understanding models use billions of common phrases and sentences to automatically learn about the world, they can also reflect human cognitive biases. Being aware of this is a good start, and the conversation around how to handle it is ongoing. Google is committed to making products that work well for everyone, and are actively researching unintended bias and mitigation strategies.

Was this article helpful?
How can we improve it?