VELORYN Documentation

Complete guide to using every feature of VELORYN — the world's first fully offline AI survival assistant. This documentation covers everything you need to know to use the app effectively, configure it for your device, and get the most out of every tool.

First Launch

When you open VELORYN for the first time, the app will run a one-time setup process. This extracts the AI models and knowledge base from the app package to your device's storage. This process takes approximately 1–3 minutes depending on your device speed.

Tip: Do not close the app during first launch setup. Let it complete fully. If interrupted, the app will retry on next launch.

After setup completes, VELORYN is ready to use. No account creation, no login, no internet connection needed — ever.

VELORYN has five main tabs accessible from the bottom navigation bar:

TabPurpose
ToolsSurvival tools, inventory management, checklist, and field journal
VoiceHands-free voice interaction with the AI assistant
Chat (center)The main AI chat interface — type questions, get survival guidance
MeshPeer-to-peer device discovery for group coordination
SettingsAppearance, device monitor, model management, advanced configuration

The Chat tab is the center tab and the primary interface. The sidebar (accessible via the menu icon on the top-left of the Chat screen) contains your chat history.

Chat Basics

Chat screen

The Chat screen is where you interact with VELORYN's AI. Type your survival question in the input bar at the bottom and tap the send button. The AI will process your question and stream a response in real-time.

How to ask effective questions

VELORYN responds best to specific, situational questions. The more context you provide, the better the response.

Less effectiveMore effective
"How to survive?""How to survive a night in a desert with no water or shelter?"
"First aid""First aid for a deep cut on my forearm that won't stop bleeding"
"Food""What plants are safe to eat in a tropical rainforest during spring?"

Stopping a response

While the AI is generating a response, the send button changes to a stop button (red square). Tap it to instantly stop the response. The AI model stays loaded and ready for your next question — stopping does not unload anything.

Read Aloud

Every AI response has a "Read Aloud" button below it. Tap it to have the response spoken through your phone's speaker. This is useful when your hands are busy or you're in a dark environment.

Environmental awareness

VELORYN uses your phone's GPS to detect your location, biome (desert, tropical, arctic, etc.), and current season. This context is automatically included in every query, so the AI tailors its responses to your actual environment. You don't need to tell the AI where you are — it already knows.

Note: GPS works fully offline via satellite signals. The first GPS lock may take 30–60 seconds without internet (cold start). Subsequent readings are instant.

Intelligence Modes

Intelligence mode selector

Tap the intelligence selector at the top center of the Chat screen to choose how the AI responds. A bottom sheet appears with four options:

ModeBest forSpeedQuality
SmartComplex decisions, medical triage, planningSlower (3–8 min)Highest
DetailedHow-to guides, step-by-step instructionsFast (30–60 sec)Good
Very DetailedLong-form guides, extensive reference materialFast (30–60 sec)Good, longest output
AutoLet the system decide based on your questionVariesOptimal for each query
Recommendation: Use Auto mode by default. It analyzes your question and picks the best mode automatically. Switch to Smart mode manually only when you need the deepest reasoning — like multi-day planning or complex medical decisions.

How Auto mode works

Auto mode analyzes the keywords and intent of your question. Emergency and medical questions are routed to Smart mode for maximum intelligence. How-to questions go to Detailed mode for speed. Planning questions get extra tokens for comprehensive output. This happens instantly with no AI processing — it's a keyword-based classifier.

Think Mode

Think Mode is a toggle in the intelligence selector bottom sheet (the lightbulb icon). When enabled, the AI reasons through your question step-by-step internally before responding. This produces higher-quality answers for complex problems but takes significantly longer.

Warning: Think Mode uses extra processing time and battery. The AI spends additional time reasoning before producing visible output. Only enable it for complex, multi-factor decisions where the quality difference matters.

When to use Think Mode

When NOT to use Think Mode

Emergency / SOS Mode

Tap the red SOS button in the top-right corner of the Chat screen to activate Emergency Mode. This changes VELORYN's behavior:

Tap SOS again to deactivate Emergency Mode and return to normal operation.

Important: Emergency Mode is designed for immediate life-threatening situations. The responses are deliberately brief. If you need comprehensive guidance, use normal mode with Smart intelligence.

Chat History & Export

Chat history sidebar

Viewing chat history

Tap the menu icon (top-left of the Chat screen) to open the sidebar. All your previous chat sessions are listed with date and time. Tap any session to reload it.

Exporting a chat

Tap the three-dot menu on any chat session in the sidebar and select Export. A popup appears with two options:

The exported file contains every message (yours and VELORYN's) with timestamps, in chronological order.

Deleting a chat

Tap the three-dot menu on any chat session and select Delete. This permanently removes that session. The action is irreversible.

Voice Assistant

Voice assistant

The Voice tab provides hands-free interaction with VELORYN. Tap the microphone button to speak your question. VELORYN converts your speech to text, processes it through the AI, and reads the response aloud.

How to use

1

Navigate to the Voice tab

2

Tap the microphone button — it will pulse to indicate listening

3

Speak your question clearly

4

VELORYN processes your speech and displays the AI response

5

The response is automatically read aloud

Note: Voice recognition uses your device's built-in speech engine. For offline operation, make sure your phone has offline speech recognition installed (Settings → Google → Voice → Offline Speech Recognition → download your language).

Survival Tools Overview

Tools overview

The Tools tab contains 20+ offline survival utilities organized as visual cards. Tap any card to open that tool. All tools work independently from the AI — you can use any tool while the AI is processing a query.

At the top of the Tools page, you'll find three buttons for productivity tools: Inventory, Tasks, and Journal. Below them is the grid of survival tools.

Compass

Compass tool

A real-time magnetic compass that uses your phone's magnetometer sensor. It shows your heading in degrees and cardinal direction (N, NE, E, SE, S, SW, W, NW).

How to use

Getting accurate readings

Warning: Magnetic north differs from true north by a few degrees (magnetic declination) depending on your location. For critical navigation, account for this offset.

SOS Flashlight

Activates your phone's camera flash in the international SOS Morse code pattern: three short flashes, three long flashes, three short flashes (··· ─── ···). This is a universally recognized distress signal.

How to use

1

Open the SOS Flashlight tool

2

Tap Start — the flash begins the SOS pattern

3

Point your phone's camera flash toward the sky or toward where rescuers might be

4

The SOS pattern repeats continuously until you tap Stop

Tip: The SOS flash continues even if you navigate away from the tool. It only stops when you explicitly tap Stop. This means you can use other features while signaling for help.

An SOS flashlight signal is visible up to several kilometers at night and can also be seen from aircraft.

Emergency Siren

Plays a loud SOS audio pattern through your phone's speaker at maximum volume. The pattern follows the SOS Morse code: three short beeps, three long beeps, three short beeps.

How to use

Warning: The siren is extremely loud. Use it only in genuine emergency situations where you need rescuers to locate you. Prolonged use will drain your battery faster.

First Aid Checklists

First aid checklists

Interactive step-by-step medical checklists for 21 conditions. Each checklist walks you through the correct procedure with checkable steps.

Available conditions

Critical emergencies: Severe Bleeding, Burns, Fractures, Snake Bite, Choking, Heat Stroke, Drowning, Allergic Reaction

Medical procedures: CPR (Adult), Checking Pulse, Recovery Position

Common conditions: Dizziness/Fainting, Eye Injury, Fever, Common Cold, Insect Sting, Migraine, Sprain/Strain, Breathing Difficulty

How to use

1

Open First Aid and tap the relevant condition

2

Read each step carefully — they are in priority order

3

Tap the checkbox on each step as you complete it

4

Completed steps are visually marked so you don't lose your place

Important: These checklists are based on established first aid protocols but are not a substitute for professional medical training. In a true emergency, always seek professional medical help if available.

Water Purification Guide

Eight purification methods with step-by-step instructions, each matched to different scenarios and available materials.

Available methods

MethodRequiresTime
BoilingFire, container1–3 minutes
Chemical TreatmentIodine or chlorine30 minutes
Cloth FiltrationClean clothImmediate
Solar Disinfection (SODIS)Clear PET bottle6 hours
Condensation CollectionPlastic sheet, containerHours
Tree Transpiration BagPlastic bag, stringHours
Rock Boiling (No Pot)Fire, rocks30+ minutes
UV Pen MethodUV sterilization device90 seconds

Each method includes what it does and doesn't remove, expected water yield, and when to use it based on your available resources.

Shelter Builder

Shelter builder guide

Visual guide to five emergency shelter types, each matched to a specific environment. Includes construction steps, materials needed, and tips for each biome.

Shelter types: Desert, Forest, Snow, Urban, and Coastal. Select the environment closest to your situation for relevant guidance.

Triage Helper

Triage helper

Based on the START (Simple Triage and Rapid Treatment) protocol used by emergency responders worldwide. This tool helps you prioritize treatment when multiple people are injured.

How it works

The tool walks you through a decision tree:

Note: Triage means making hard decisions about who to help first. The goal is to save the most lives possible, not to treat the most severely injured first. Follow the protocol — it's designed for exactly these situations.

Unit Converter

Unit converter

A comprehensive offline converter supporting six categories:

How to use

Blood Type Compatibility

Blood type checker

Select your blood type (A+, A-, B+, B-, AB+, AB-, O+, O-) to instantly see:

A general blood safety warning is always displayed regardless of which type is selected. Specific compatibility information appears above the general warning when you select a type.

Critical: Blood transfusions should only be performed by trained medical professionals. This tool is a reference for emergency situations where knowing compatibility could save a life while waiting for professional help.

Sun Position Calculator

Sun position calculator

Calculates sunrise and sunset times based on your GPS coordinates and the current date. No internet needed — uses astronomical formulas.

Useful for planning travel (avoid moving at night), shelter orientation (face openings away from the setting sun for warmth), and estimating remaining daylight hours.

Altitude / Barometer

Reads your phone's pressure sensor to estimate altitude above sea level and display barometric pressure.

Note: Altitude estimation from pressure is approximate (±50m). Weather changes affect pressure readings. For critical altitude needs, cross-reference with terrain features.

Morse Code Translator

Type any text and see the Morse code equivalent in dots and dashes. Useful for signaling with a flashlight, tapping on metal, or any on/off signaling method.

Common codes to know: SOS = ··· ─── ···

Timer / Countdown

A countdown timer with custom time input. Set any duration and the timer counts down with audio notification when complete.

Key feature

The timer persists across navigation — if you start a timer and switch to another tool or the chat, the timer keeps running in the background. When it completes, you'll be notified.

Useful for: water purification timing (boiling, chemical treatment wait times), medication intervals, shift rotations in group survival, and rest periods.

World Clock

World clock

Shows the current time in major cities around the world. Fully offline — calculated from UTC offset and your device clock.

Useful when coordinating with rescue teams, embassies, or contacts in different time zones.

Signal Mirror Angle Calculator

Uses your compass heading and the sun's position to calculate the exact angle you need to aim a signal mirror to reflect sunlight toward a target (aircraft, distant rescuers).

Signal mirrors are one of the most effective long-range rescue signals — effective up to 50km in clear conditions — but only if aimed correctly. This calculator removes the guesswork.

Rain Collector Calculator

Input your container size and estimated rainfall rate to calculate how long it will take to collect a drinkable amount of water. Helps with water planning and managing expectations during rain events.

Body Temperature Estimator

Estimates core body temperature using a symptom checklist and pulse rate input — no thermometer needed.

Warning: This is an estimation tool, not a medical device. Use it to identify trends and make informed decisions. Always err on the side of caution — if in doubt, treat for the suspected condition.

Edible Plant Quick Reference

Text-based identification cards for common edible plants organized by biome. Each card includes physical description, habitat, seasonal availability, preparation methods, and warnings about poisonous look-alikes.

Critical: Never eat any plant you cannot identify with 100% certainty. When in doubt, don't eat it. One wrong plant can cause organ failure and death. Use the Universal Edibility Test if you must test an unknown plant.

Inventory Management System

Inventory bento view

A complete resource tracking system for survival situations. Track what you have, what you're consuming, and predict when supplies will run out.

Creating categories

1

Tap the Inventory button at the top of the Tools page

2

Tap the + button

3

Select an existing category from the dropdown, or tap New Category to create one (e.g., Food, Water, Medicine, Tools)

4

Enter the item name, quantity, and unit

5

Tap Save

Add inventory item

Updating quantities

Tap the Edit button on any item card to update its quantity. Every update is logged with a timestamp.

Viewing history

Inventory history view

Tap any item card or the History button to see the complete history of quantity changes over time. This shows your consumption rate and helps you estimate remaining supply days.

Stock indicators

Each item card shows an arrow indicator: an up arrow (green) means stock increased since last update, a down arrow (red) means stock decreased. This gives you an at-a-glance view of your supply trends.

Filtering

Use the category filter to show only items from a specific category (e.g., show only Food items).

Deleting categories

Long-press or use the delete option on a category. A confirmation dialog will appear: "Do you want to delete [category name]? This action is reversible." Choose Yes, delete or No, keep it.

Tasks / Checklist

Checklist tool

A simple todo list for managing survival tasks.

Use cases: daily water collection assignments, scouting routes to check, group task delegation, supply gathering lists, shelter maintenance tasks.

Field Journal

A full-featured note-taking tool designed for survival documentation.

Creating a note

1

Tap the Journal button at the top of the Tools page

2

Tap the + button in the center

3

A full-screen text editor opens — start typing

4

Your note is auto-saved every 3 seconds and when you press back

Titles

If you don't set a title, VELORYN automatically uses the first sentence of your note as the title. If the first sentence is too long, it's truncated with "..." You can edit the title at any time.

Color coding

Assign one of 10 colors to any note for visual organization. The title card in the journal home page displays in your chosen color. Use colors to categorize: red for urgent, blue for water-related, green for food, etc.

Journal home view

Notes are displayed as a bento box grid showing the title and date/time. Tap any card to open and edit. Use the 3-dot menu button on each card's top-right corner for options: Edit, Share, Delete.

Sharing notes

Tap the 3-dot menu → Share. This uses your device's native share function, allowing you to send the note via any available method (nearby share, Bluetooth, etc.).

Deleting notes

A confirmation dialog appears before deletion to prevent accidental loss.

Mesh Network / Device Discovery

Mesh network

The Mesh tab enables peer-to-peer device discovery using Bluetooth and WiFi Direct. This works without cell towers, internet, or any infrastructure.

How to use

1

Navigate to the Mesh tab

2

Tap Scan for Devices

3

Nearby devices running VELORYN (within ~100 meters) will appear in the list

This feature is designed for disaster scenarios where groups need to coordinate without cellular infrastructure. Ensure Bluetooth and WiFi are enabled on your device.

Settings: Appearance

Color Mode

Toggle between Light and Dark themes. Your preference is saved and persists across app restarts.

Font Style

Choose between System font (your device's default) or VELORYN style (the app's custom font). Default is VELORYN style.

Font Size

Adjust the text size across the entire app. Options: Small, Medium, Large (default), Extra Large. Larger sizes are recommended for stressful situations where readability under duress matters.

Your font size preference is saved and persists across app restarts.

Settings: Resource Monitor

Resource monitor

The Resource Monitor is at the top of the Settings screen and shows real-time device information:

MetricWhat it shows
Free RAMAvailable memory in megabytes. The AI model needs at least 600MB free to load.
BatteryCurrent battery percentage.
Battery TemperatureInternal battery temperature. Above 45°C indicates the device is running hot.
Device TemperatureSkin/CPU thermal reading. Above 50°C means the device is severely throttling.
AI EngineShows "Ready" (green) if a model is loaded, or "Not ready" (grey) if no model is loaded.
Knowledge BaseNumber of indexed documents from the survival database.
GPSCurrent coordinates, detected biome, and season.

Refresh button

Tap the reload icon on the right side of the Resource Monitor header to instantly refresh all readings. The monitor also updates automatically at regular intervals.

Tip: If the device temperature is consistently above 45°C, unload the AI model when not in use. Extended high temperatures can damage your battery and reduce your phone's lifespan — critical when your phone is your only survival tool.

Settings: Model Management

Unload Model

The Unload Model button immediately frees the AI model from RAM. The button is located above the Advanced section in Settings.

When you tap Unload, a confirmation dialog appears:

Benefits of unloading:

The model will automatically reload when you send your next chat query.

Note: All tools, inventory, journal, and other features continue to work normally after unloading the model. Only the AI chat and voice features require a loaded model.

Settings: Survival Mode

A toggle to switch between normal mode and survival/emergency mode globally. When enabled, all AI responses are optimized for life-critical situations — shorter, more action-oriented, and focused on immediate survival priorities.

This is the same as tapping the SOS button in the Chat screen, but accessible from Settings for convenience.

Settings: Advanced

Advanced settings
Warning: The Advanced settings section contains controls that can affect AI performance and stability. We do not recommend changing these settings unless you understand what each option does. Incorrect values may cause degraded responses, excessive battery drain, or app instability.

Response Length

Controls the maximum number of tokens (roughly words) the AI can generate per response.

SettingEffect
Low (100–300)Very short responses. Suitable for low-end devices or quick questions.
Default (800)Balanced. Recommended for most devices and use cases.
High (1000–2000)Long, detailed responses. Requires more processing time and battery.
Maximum (3500)Extremely long output. Only for powerful devices with 8GB+ RAM. Will cause significant heat generation.

Two quick buttons are available:

Consequence of setting too high: If the response length exceeds what your device can handle, the AI may produce garbled output, extremely slow responses, or the app may become unresponsive. If this happens, reduce the value or tap "Default."

CPU Threads

Controls how many CPU cores the AI uses for inference.

SettingEffect
1 threadSlowest but coolest. Minimal heat generation. Good for preserving battery.
2 threads (Default)Balanced. Uses the two fastest CPU cores. Recommended for most phones.
4 threadsFaster prefill but may cause thermal throttling on mid-range devices.
6–8 threadsMaximum speed. Only for flagship devices with 4+ big CPU cores. Will cause significant heat.
Consequence of setting too high: Using more threads than your device has high-performance CPU cores will actually slow down the AI (weak cores bottleneck fast ones) and cause severe overheating. If your phone becomes hot to the touch, reduce threads immediately.

Auto-Unload on Inactivity

Automatically unloads the AI model after a period of inactivity to save battery and reduce heat.

SettingBehavior
1 minute (Default)Model unloads 1 minute after your last query. Good for casual use.
2, 5, 10, 30 minutesLonger timeouts for sessions with intermittent queries.
NeverModel stays loaded until you manually unload it or close the app. Maximum convenience but highest battery drain.
Note: The auto-unload timer will never fire during an active generation. If the AI is mid-response when the timer expires, it waits until the response is complete before unloading.

Knowledge Retrieval Sensitivity

Controls how strictly the knowledge base filters search results before feeding them to the AI.

SettingEffect
OffAll retrieved chunks are included regardless of relevance. May confuse the AI with irrelevant data.
Low (0.3)Most chunks included. Higher chance of finding relevant information, but also more noise.
Medium (0.4)Balanced retrieval.
Default (0.5)Recommended. Good balance between finding relevant information and filtering noise.
High (0.6)Stricter filtering. Only highly relevant chunks are included. May miss some useful information.
Very High (0.7)Very strict. Only near-exact matches are included. The AI will rely mostly on its own knowledge.
Consequence of changing: Setting too low floods the AI with irrelevant text, producing garbled or off-topic responses. Setting too high means the AI gets no reference material and must answer entirely from general knowledge, which may lack specific survival details. A confirmation dialog appears when changing from the default value.

View Logs

Opens the diagnostic log viewer. Logs are written to disk and survive app crashes. Each log session shows a timestamp and contains entries for every significant event (service initialization, model loading, query processing, errors).

Logs are useful for troubleshooting issues. If you experience a crash or unexpected behavior, check the logs immediately after — they contain the exact sequence of events that led to the issue.

Understanding Log Entries

Log entries follow this format: [time] [severity] message

SeverityMeaning
INFONormal operation — service started, model loaded, query processed
WARNSomething unexpected but non-critical — fallback search used, partial result
ERRORSomething failed — model failed to load, generation error
FATALCritical failure — usually the last entry before a crash

When reporting issues, include the full log session from the time the problem occurred.

Troubleshooting

Most likely cause: Low available RAM. The AI model needs at least 600MB of free RAM to operate.

Solution:

  • Close other apps running in the background
  • Check the Resource Monitor in Settings — if Free RAM is below 600MB, close background apps
  • If the issue persists, try switching to Detailed or Very Detailed mode (uses a smaller model)
  • Reduce the Response Length in Advanced Settings

Cause: The AI model started generating but produced no output. This can happen when the model was auto-unloaded during generation or when the prompt exceeds the context window.

Solution:

  • Try sending the same message again
  • If the auto-unload timer is set to 1 minute, increase it to 5 or 10 minutes
  • Reduce the Response Length if set above 1000
  • Clear the chat and start a new conversation (old messages may be filling the context)

Cause: The knowledge retrieval returned irrelevant chunks that confused the AI, or the Response Length is set higher than the context window can support.

Solution:

  • Set Knowledge Retrieval Sensitivity to Default (0.5) in Advanced Settings
  • Reduce Response Length to 800 (Default)
  • Try rephrasing your question with more specific terms
  • Switch to Smart mode for the highest quality responses

Cause: AI inference is computationally intensive. The CPU is working at high capacity to process your queries.

Solution:

  • Reduce CPU Threads to 1 or 2 in Advanced Settings
  • Enable auto-unload (1 or 2 minutes) so the model unloads between queries
  • Manually unload the model when not actively asking questions
  • Use Detailed mode instead of Smart mode — it uses a lighter model that generates less heat
  • Remove your phone case to improve heat dissipation
  • Avoid using VELORYN while charging — charging + AI inference generates extreme heat

Monitor: Check the Resource Monitor. Battery temp above 45°C or Device temp above 50°C means your device is throttling and may become unstable.

Cause: In earlier versions, stopping generation could cause a conflict. This has been fixed in recent updates.

Solution:

  • Make sure you're running the latest version of VELORYN
  • If the issue persists, wait for the response to complete rather than stopping it
  • Check the logs for details and report the issue

Cause: Your device may not have offline speech recognition installed. Without it, the speech engine times out immediately.

Solution:

  • Go to your phone's Settings → Google → Voice → Offline Speech Recognition
  • Download the English (or your preferred language) speech pack
  • This requires a one-time internet connection to download (~50MB)
  • After downloading, voice input works fully offline

Cause: GPS requires clear sky visibility. Indoor use, dense forest canopy, or urban canyons can block satellite signals.

Solution:

  • Move to an open area with clear sky view
  • Wait 30–60 seconds for the initial GPS fix (cold start takes longer without internet)
  • Ensure location permissions are granted to VELORYN in your device settings
  • The GPS icon in the status bar should be active when VELORYN is reading location

Cause: Magnetic interference from nearby metal, electronics, or magnets.

Solution:

  • Calibrate: move your phone in a figure-8 pattern several times
  • Move away from metal objects, vehicles, speakers, and other electronics
  • Hold the phone flat and level — tilting causes inaccuracy
  • Some phone cases with magnetic clasps can interfere — try removing the case

Explanation: Battery temperature changes slowly (1–2°C per minute) which can make it seem stuck. Device temperature is more responsive but depends on your device's thermal sensors.

Solution:

  • Tap the refresh button in the Resource Monitor to update readings
  • Wait 30 seconds between refreshes for meaningful changes
  • The readings are accurate — temperature simply doesn't change as fast as other metrics

Expected behavior: Both the SOS Flashlight and Emergency Siren should continue running when you navigate away. They only stop when you explicitly tap Stop.

If they're stopping unexpectedly:

  • Make sure you're running the latest version
  • Some devices aggressively kill background processes — check your device's battery optimization settings and exempt VELORYN
  • Go to Device Settings → Apps → VELORYN → Battery → set to "Unrestricted"

Cause: Your device may be aggressively clearing app data or the app storage might be corrupted.

Solution:

  • Make sure VELORYN is not set to "Clear data on close" in your device settings
  • Exempt VELORYN from any "battery optimizer" or "memory cleaner" apps
  • Check that your device has at least 100MB of free internal storage

Performance Tips

Maximizing battery life

Getting the best AI responses

Keeping your phone alive as long as possible

Remember: In a survival situation, your phone's battery is your lifeline. Every minute of screen time, every AI query, every flashlight pulse costs power. Use VELORYN efficiently — get the information you need, then act on it.

VELORYN Documentation · © 2026 AegisVectro · Back to Home