Speak naturally
Hold the input field to talk. Release to finish, or slide up and release to cancel.
YOUR ANDROID, A LITTLE MORE HELPFUL
Say what you need.
Let your phone help.
A conversation that can turn into action. Speak or type, add a photo, and let Tapilot work with the apps on your phone.
9:41
TapilotILLUSTRATIVE CONVERSATION
Help me organize the information on this screen.
TapilotI can work with what’s on your screen. Tell me what you’d like to keep.
Hold to speak
Optional voice polishing
01 / TAPILOT
Made for the moment.
From a passing thought to a task you can follow.
Keep the thread
A dedicated conversation agent keeps the context and hands phone actions to Midscene.
Bring more context
Attach a photo, take a picture, or share a file with your assistant.
See what happened
Open the execution report to review actions, screenshots, and failures.
02 / GET STARTED
Set it up once. Then just ask.
- 1
Connect your model
Bring your own compatible model API. Configure speech recognition separately with Volcengine or Deepgram.
- 2
Choose the access
Enable accessibility for phone actions. Screen capture and microphone access are requested when needed.
- 3
Start a conversation
Talk from the chat or global assistant panel. Review the conversation and execution reports as you go.
Optional voice polishing
Remove filler words and organize a transcript with a separately configured model. If polishing fails, Tapilot keeps the recognized text.
Your services. Your choice.
Use your own API credentials. Keys are encrypted using Android Keystore. Model, speech, and optional polishing requests go to the providers you configure. Reports stay in app-private storage until you share them.
Privacy & permissions →A small assistant for the things you want to do.
Download Android beta ↗Beta software. App compatibility varies. Requires your own API services; provider fees may apply.