Connect cloud AI through Google Gemini or OpenAI-compatible APIs, including compatible providers such as OpenRouter and Anthropic.
Bring your own API keys and keep them protected using the Android Keystore.
Local and cloud AI are intentionally different within Vulcan. Local models can access Vulcan's deeper environment, while cloud models do not receive direct access to the Archive Network or Smart Reflexes.
Matrix Learning allows supported local models to work with information you provide.
Paste text, upload PDF or text files, or fetch supported web content such as Wikipedia pages.
When Matrix Mode is enabled, the local AI answers from the selected material as its source of truth, creating a focused environment for working with specific documents or knowledge.
When you're ready to make that knowledge part of the AI's broader knowledge base, use Learning to summarize, process and add the information to the Archive.
This means information can begin as a temporary, focused source in Matrix Mode and become persistent knowledge available to your local AI outside of Matrix Mode.
The Archive Network is Vulcan's persistent knowledge system for local AI.
Instead of treating information as an undifferentiated memory store, Vulcan organizes knowledge into drawers, folders and records, creating a structured archive that can be searched when relevant information is needed.
The Archive supports:
Information learned through Matrix Learning can become part of this broader knowledge base, allowing your local AI to retrieve it later when it becomes relevant.
Vulcan can allow supported local models to observe conversations and extract useful information in the background.
Observations can identify:
Personal information
Knowledge topics
Categorized subjects
AI-generated insights
These observations contribute to Vulcan's Familiarity system, allowing the AI environment to gradually build a richer understanding of the user's context.
Observations are controlled by the user and are only available to supported local models.
Smart Reflexes give supported local AI the ability to interact with Android through natural language.
Instead of simply telling you what to do, the AI can perform supported actions such as:
Control media playback
Turn the flashlight on or off
Make phone calls
Send SMS
Draft emails
Navigate with Google Maps
Create and manage alarms
Set and control timers
Adjust volume
Control Wi-Fi and Bluetooth
Toggle Do Not Disturb
Launch applications
Open Android settings
Read calendar events
Create calendar events from natural language
Stop active actions
Reflexes also understand contextual follow-ups. A command such as "turn it off" can refer to the device action performed immediately before it.
Smart Reflexes are part of Vulcan's local AI environment. Cloud models do not receive direct access to this device-control layer.
API Uplink allows Vulcan to connect AI conversations to external REST APIs.
Configure endpoints, variables and authentication methods including:
Bearer authentication
Basic authentication
API keys
Once configured, AI can invoke supported APIs through the call_api capability, allowing Vulcan to interact with external services beyond its built-in tools.
Scheduled Prompts turn AI into a scheduled assistant.
Create prompts that run:
Once
Daily
Weekly
Monthly
Yearly
Vulcan can deliver the resulting AI response as an Android notification instead of simply sounding an alarm.
Recurring prompts can be paused and resumed, and Vulcan can prepare varied response templates so scheduled interactions don't have to feel identical every time.
Vulcan provides a full AI conversation environment with:
Streaming responses
Image and video attachments
Speech-to-text
Text-to-speech
Automatic speech playback
Multiple TTS engines
Thinking and generation stages
Response regeneration
Generation controls
Performance statistics
Message feedback
Session search and management
AI responses can be streamed token by token, with information such as time-to-first-token and tokens-per-second available when supported.
Vulcan can also generate images through supported cloud providers, with generated images stored locally.
The Deck is Vulcan's device-awareness and diagnostics environment.
At its center is an interactive representation of the device. Each subsystem can be explored by selecting its corresponding block, opening detailed information relevant to that part of the device.
The Deck provides access to information including:
CPU
GPU
RAM
Storage
Battery
Network
Thermal state
Camera
NFC
Bluetooth
Security information
Device configuration
Permissions
Sensor information
Vulcan exposes a broad collection of sensor tools covering spatial, inertial, magnetic, atmospheric, audio, RF and power measurements.
A persistent notification can also provide live CPU, RAM, GPU and device temperature readings while Vulcan is running.
The Deck isn't simply a system-information page. It provides Vulcan with awareness of the device it operates within.
Define How Your AI Behaves
Directives allow you to define persistent behavioral rules for your AI.
Create, edit, list and remove directives directly through chat, allowing you to establish how your AI should respond or behave without repeatedly explaining the same instruction.
Directives are protected by Vulcan's safety guardrails during creation.
Vulcan is designed around user control and privacy.
When using local AI, processing can remain entirely on the device. Local models can use Vulcan's Archive Network, Smart Reflexes and other integrated capabilities without sending that information to a cloud AI provider.
When using cloud providers, Vulcan connects through the provider's API according to the configuration you choose. Cloud models do not receive direct access to Vulcan's local Archive Network or Smart Reflexes.
API keys protected with Android Keystore
No analytics
No crash reporting
No tracking
Android backup disabled for application data
Safety guardrails across generation paths
Additional safety controls for supported cloud providers
Vulcan also includes safeguards around device actions to reduce the risk of an AI misinterpreting a conversational statement as a command.
One Environment. Many Capabilities.
Vulcan brings together capabilities that are normally separated across different applications and services.
AI models provide intelligence.
Matrix Learning provides focused knowledge.
The Archive Network provides persistent searchable knowledge.
Observations build contextual familiarity.
Smart Reflexes let local AI act.
API Uplink connects AI to external services.
Scheduled Prompts let AI operate on a schedule.
The Deck gives Vulcan awareness of the device it runs on.
Directives let you define how your AI behaves.
Together, these systems turn Vulcan from a conventional AI chat application into something broader:
An AI Operating Environment.
Run your AI locally. Connect it to the cloud. Give it knowledge. Let it learn. Let it remember. Give it tools. Let it act.
Vulcan: AI Hub.