Installation and user guide.
Learn how to install OTO AI, connect an AI provider, prepare meeting context and use the live meeting tools.
OTO AI can assist with interviews, collaborative discussions, science, software, healthcare, research, planning and other subjects. It can listen for questions, provide contextual answers and analyse visible screen content when a compatible model is selected.
Install OTO AI
Download the installer supplied by Sanity OS or your authorised administrator.
- 1
On Windows, open the OTO AI installer file and review the licence agreement.
- 2
Accept the agreement and follow the installer instructions.
- 3
On macOS, open the disk image and move OTO AI into Applications.
- 4
Open OTO AI and approve the microphone and audio permissions requested by your operating system.
If macOS blocks the first launch, open System Settings, choose Privacy & Security, then select Open Anyway.
Welcome to the OTO AI setup.
Connect an AI provider
OTO sends questions and optional meeting context directly to the provider you select. You need your own API key.
- 1
Open Settings and select AI models.
- 2
Select Add model, then choose a provider and model.
- 3
Paste your API key and select Test connection.
- 4
Once the connection succeeds, save the model and make it active.
Saved keys are held in your operating system credential store. Treat every API key like a password.
Connect an AI provider
Where to create a key
Provider accounts, usage limits and charges are managed separately by each provider.
Sign in to the API platform, create a secret key and confirm that API billing is enabled.
Open API keys ↗Sign in to the Claude Platform, add credits if required and create a key under API Keys.
Open API keys ↗Use Google AI Studio to create a key connected to your Google Cloud project.
Open API keys ↗Sign in to the DeepSeek platform, add account credit if required and create an API key.
Open API keys ↗Choose a meeting mode
The home screen gives you three ways to begin.
- ≋
Go with the flow starts immediately without prepared context. Use it when the conversation itself provides enough information.
- ＋
Provide context lets you describe the meeting and add documents or public web references before starting.
- ↺
Previous subjects lets you use, edit or remove context saved from an earlier meeting.
Provide meeting context
Prepared context helps the selected model understand the subject before questions are asked.
- 1
Give the meeting a short, recognisable name.
- 2
Describe the meeting type, participants, goals, terminology and desired outcome.
- 3
Add TXT, Markdown, DOCX or PDF files when they contain useful reference material.
- 4
Add public web pages that OTO should use as references.
- 5
Select Save & start meeting.
Provide meeting context
Use the live session
The compact overlay keeps the transcript and answer controls available without covering your meeting.
- 1
Use the listening control to start or stop capture from calls and media.
- 2
Use the microphone control to start or stop OTO listening to you. This does not mute your microphone in other applications.
- 3
Read the recognised question in the Live Transcript bar.
- 4
Open AI Answers to see responses, details and key points.
- 5
New answers join the queue without moving the answer you are currently reading.
Analyse your screen
Screen analysis helps explain a visible diagram, chart, slide, document or application.
- 1
Choose an AI model that supports image input.
- 2
Enable Analyze screen for that model.
- 3
Display the material you want the model to inspect.
- 4
Use the screen analysis control or its configured hotkey.
- 5
Wait for the progress message, then review the answer and key points.
Review confidential information before analysing your screen. Visible content is sent to your selected AI provider.
Settings reference
Settings remain fully visible even when the meeting overlay uses reduced opacity.
Add, edit, test, select or disable AI provider connections.
Download and view the local speech recognition model.
Choose the speaker output and microphone input used by OTO.
Control answer expansion, casual question filtering, answer detail, queue size, capture pausing and question timing.
Choose light, dark or system mode. Set overlay opacity and answer text size.
Set shortcuts for visibility, sessions, microphone capture, answers and screen analysis.
Avoid common operating system shortcuts such as Ctrl + A. Use Reset hotkeys if a shortcut causes a conflict.
Privacy and data
OTO is designed to keep local data local while sending only necessary requests to your chosen AI provider.
Speech recognition runs on your device.
Saved API keys use the operating system credential store.
Questions, selected context and requested screen images go directly to your chosen provider.
Content protection is best effort. Results can vary by operating system, recording software and remote desktop tool. Always follow applicable consent rules and meeting policies.
Frequently asked questions
Why can I not start a meeting?⌄
Open AI models in Settings. Add an API key, test the connection, save the model and select it as the active model.
Does microphone mute affect other applications?⌄
No. The microphone button controls only whether OTO listens to your microphone. It does not mute you in the meeting application.
Why is Analyze Screen not available?⌄
The control is available only when the active model supports image input and screen analysis is enabled for that model.
Where are API keys stored?⌄
Saved keys are stored in your operating system credential store. They are not included in the OTO AI installer.
How do I reopen OTO after closing it?⌄
Closing OTO hides the application. Use your configured Show or Hide hotkey to reopen it. Use Quit OTO AI in Preferences when you want to exit completely.
Can OTO guarantee that nobody can capture the overlay?⌄
No. OTO requests operating system content protection, but protection can vary between operating systems, screen capture tools and remote desktop products.