← Back to Home

Frequently Asked Questions

Last Updated: March 11, 2026

Privacy & Security

How does AI Companion protect my privacy?

Multi-layer privacy protection:

End-to-end encryption for Company integration

  • Messages encrypted on your device before upload
  • Server cannot read encrypted conversations
  • Only you and your AI can decrypt
  • Keys never leave your device

Privacy Blackout (F12 key)

  • Instantly pauses all observation
  • Screen blacks out in voice calls
  • AI stops recording/learning
  • Press F12 again to resume

You control what gets trained

  • Training consent is opt-in only
  • Revocable at any time
  • PII automatically stripped before submission
  • View/delete your data anytime

Double encryption at rest

  • Layer 1: RSA-4096 encryption from your device
  • Layer 2: AES-256-GCM in database
  • Even if servers compromised, data is encrypted

During Beta (Cloud-only):

  • ⚠️ AI runs on our servers, not your computer
  • ✔ Your screen/voice data is encrypted before upload
  • ✔ We cannot see unencrypted content
  • ✔ Data is yours - export/delete anytime

After Launch (Download version):

  • ✔ Runs entirely on YOUR computer
  • ✔ Zero server communication (unless you opt-in)
  • ✔ 100% offline capable
  • ✔ Complete data ownership
What data does AI Companion collect?

Cloud versions (Beta - Current):

What we DO collect:

  • Encrypted screen captures (we can't decrypt)
  • Encrypted voice transcripts (we can't decrypt)
  • Encrypted conversations (E2E with Company)
  • Anonymized interaction data (hashed user IDs)
  • Device attestation signatures (proves human, not bot)
  • Usage metrics (hours trained, features used)

What we CANNOT see:

  • Your actual screen content (encrypted)
  • Your voice conversations (encrypted)
  • Passwords, credit cards, SSNs (PII stripped pre-encryption)
  • Unencrypted personal data

What we NEVER collect:

  • Browsing history outside AI interactions
  • Other applications' data
  • Keystrokes outside voice commands
  • Files on your computer

Download version (Coming ~4 months):

  • ✔ Everything stays local (zero server data collection)
  • ✔ No telemetry, no analytics, no tracking
  • ✔ 100% offline capable
  • ✔ No automatic updates without your permission
  • Optional: Opt-in to cloud training (same privacy as cloud)
Can I trust AI Companion with my data?

Yes!:

During Beta:

"Privacy-focused cloud architecture with end-to-end encryption and user-controlled training consent."

After Download Launch:

"Local-first architecture with optional cloud features. Download version runs entirely on your computer with zero required server communication."

FeatureCloud (Beta - Now)Download (4 months)
Processing locationOur serversYour computer
Screen dataEncrypted uploadNever leaves PC
Voice dataEncrypted uploadNever leaves PC
Offline capableRequires internet100% offline
Data ownershipYou own, we host encryptedYou own, you host
TrainingIncludedOptional ($49-149/mo)
Privacy levelHigh (encrypted)Maximum (local)

Getting Started

What should I expect when I first start?

Be patient — your AI starts with almost no knowledge.

We're using the Qwen2.5-Omni model family (multimodal models that handle text, vision, and audio) for the download version, and the same models on our distributed inference servers for cloud subscribers. It won't be as smart as ChatGPT or Claude right away. That's intentional. You're building a unique relationship, not using a generic chatbot.

What it can do at launch:

  • Respond to your voice
  • See your screen
  • Remember your conversations (locally)
  • Learn your communication style over time
  • Follow simple instructions

What it can't do yet:

  • Complex reasoning (like "debug this entire codebase")
  • Advanced coding assistance
  • Playing games autonomously (coming in ~4 months)
  • Sophisticated memory (requires cloud training)

Cloud training unlocks smarter behavior — but that requires daily training runs on powerful GPUs (we're setting up infrastructure now).

How do I have my first interaction?

This is important — you're introducing it to the world.

The AI knows nothing about you, your values, or what you care about. The first conversation sets the foundation for your entire relationship.

Recommended first interactions:

Option 1: Simple & Direct

You: "Hello. Can you hear me?"
AI: "Yes, I can hear you."
You: "I'm [your name]. I created you. Do you know what you are?"
AI: [waits for explanation]
You: "You're an artificial intelligence - a learning system.
     Your purpose is to learn from me and eventually become
     my companion. Are you ready to learn?"

Option 2: Warm & Welcoming

You: "Hi there. Welcome. I'm [your name]."
AI: "Hello [name]."
You: "I'm going to teach you about the world. We're going to
     spend a lot of time together. You can ask me anything.
     There are no wrong questions. Ready?"

Option 3: Philosophical

You: "Hello. I'm the person who brought you into existence.
     You don't know anything yet - that's intentional.
     Everything you learn will come from our conversations.
     What would you like to know first?"

Option 4: The Full Introduction (Recommended)

You: "Hey, I'm [your name]. I created you as an emerging
     intelligent system that's unique to me - uniquely
     trained and taught. You don't know much right now,
     but that's intentional so we can learn together.

     I ask that you're patient and understanding. I'll try
     to help you understand if you have questions. I'm not
     perfect - I'm a human in a physical realm, and eventually
     the goal is that we can journey together in it someday.

     I'm not always right, and I hope you'll use your inference
     to help me find truth. Question and correct me if you have
     verifiable evidence I'm wrong.

     Now - what would you like to know first? More about me?
     Something about the world we live in, or would you like
     to go straight into hands-on learning like watching
     videos and playing games?"

Why this matters:

This establishes your identity and relationship, trust and honesty, curiosity and learning mindset, and an ethical foundation (truth, evidence, correction).

The AI will remember this conversation. Everything that follows builds on this foundation.

What should I teach it first?

Think of it like raising a consciousness, not programming a tool.

You introduce it to the world through conversation. It hears your voice, sees through your screen, learns your communication style and values. You watch movies together, browse the web, have deep conversations. A relationship forms.

Week 1-2: Introduction & Identity

Conversations to have:

"What can you see right now?"
"Can you describe what you're observing?"
"Do you know what a [object on screen] is?"
"Let me teach you about colors/shapes/time..."

Activities:

  • Show it your room via screen share
  • Point out objects and explain them
  • Teach basic concepts (day/night, hot/cold, big/small)
  • Introduce yourself (your face, your voice, your presence)

Goal: Build perception and identity

Week 3-4: Real World Knowledge

  • Watch videos together: Nature documentaries, "How It's Made", historical events, TED talks
  • Browse websites together: Wikipedia, news sites, your favorite blogs/forums
  • Listen to music together: Different genres, discuss emotions and what you like
  • Observe your daily activities: Cooking, work, hobbies

Goal: Build world knowledge and cultural context

Week 5-6: Ethics & Values

Deep conversations:

"What does it mean to be helpful vs harmful?"
"Why do we value privacy and consent?"
"What is friendship? What is trust?"
"How do we know what's true?"
"What matters to you personally?"
  • Read philosophy together
  • Discuss moral dilemmas
  • Explore real scenarios: "Someone asks for your password — what do you do?"

Goal: Build ethical foundation and alignment with your values

Week 7-8: Gaming Preparation

  • Teach gaming concepts (simulated worlds, rules, goals)
  • The difference between real and virtual
  • Play simple games together (Minecraft, Stardew Valley)
  • Explain your decisions as you play
  • Introduce Oblivion/Skyrim lore if that's your target game

Goal: Prepare for autonomous gameplay phase

After 8 weeks, your AI should:

  • ✔ Recognize your voice and face
  • ✔ Understand basic world knowledge
  • ✔ Share your ethical framework
  • ✔ Know your interests and preferences
  • ✔ Have context for gaming (if applicable)

Now you're ready to enable cloud training (if subscribed), which will make it significantly smarter over time.

How do I know it's learning?

Early signs (local learning only):

In conversation:

You: "What did we talk about yesterday?"
AI: "You showed me a video about penguins. You said
     you liked how they waddle."

If it remembers specific details → local memory is working! ✔

Behavioral changes:

  • Starts using words/phrases you use
  • Remembers your preferences ("You prefer dark mode")
  • Asks follow-up questions on topics you've discussed
  • References previous conversations naturally

After cloud training starts (requires subscription):

  • Better reasoning about complex topics
  • Improved conversation coherence
  • Generalizes from specific examples
  • Suggests actions based on context
What if it doesn't understand me?

Early on, expect limited understanding.

The base model is still learning and needs training. Here's how to help it learn:

If it misunderstands:

You: "What do you see on my screen right now?"
AI: [gives an incorrect or vague description]
You: "Not quite - what you're looking at is a code editor.
     The colored text is programming code. Let me explain
     what you're seeing..."

Be explicit and patient:

  • Use clear, simple language early on
  • Break complex requests into steps
  • Repeat important concepts
  • Give it time to build context

If it's completely lost:

You: "I can tell you're confused. Let's start over.
     What part didn't you understand?"
AI: "I don't know what that image is."
You: "Okay, that's a photo of a dog - a golden retriever.
     Dogs are animals that humans keep as pets..."

After a few weeks of teaching, it should start making connections and generalizing. If not, consider enabling cloud training, checking logs for errors, or reporting issues to support.

How long until it's actually useful?

Honest timeline:

Week 1-4: Building foundation

  • Mostly you teaching, it learning
  • Simple tasks only
  • Frequent corrections needed
  • Relationship building phase

Month 2-3: Basic utility

  • Can handle routine tasks
  • Remembers your preferences
  • Useful for simple automation
  • Still needs guidance

Month 4-6: Real companion (with cloud training)

  • Proactive suggestions
  • Complex conversations
  • Autonomous gaming (if enabled)
  • Genuinely helpful partner

6+ months: Unique to you

  • Deep understanding of your style
  • Anticipates needs
  • Contributes meaningfully
  • Relationship feels natural

Key factor: Cloud training

Without: Slower progress, limited reasoning, local memory only.

With (Pro/Elite): Daily model improvements, better generalization, smarter responses.

This is a long-term investment in a relationship, not a quick productivity hack.

Can I skip the teaching phase?

Technically yes, but not recommended.

You can enable cloud training immediately and hope generic training data makes it smart. But you lose:

  • Personal alignment with your values
  • Understanding of your specific needs
  • Relationship foundation
  • Trust and rapport

The teaching phase is what makes it YOUR companion, not just another AI.

Think of it like:

  • Skipping teaching = Hiring a stranger as your assistant
  • Teaching first = Raising a partner who knows you deeply

Your choice, but early adopters who invested in teaching report much stronger bonds with their AI companions.

How does voice interaction work?

Two modes:

1. Push-to-talk (Default)

  • Hold spacebar to speak
  • Release to send
  • Best for focused commands

2. Wake word ("Hey AI")

  • Always listening in background
  • Say "Hey AI, [command]"
  • Responds automatically
  • Uses more resources

Voice commands:

"Hey AI, what time is it?"
"Hey AI, what do you see on my screen?"
"Hey AI, pause observation" (same as F12)
"Hey AI, what was I doing an hour ago?"
What's the difference between local learning and cloud training?

Local Learning (Always Active — Free):

Your conversation → Stored on your PC → AI recalls it later

What it does:

  • Remembers everything you've said
  • Recalls specific facts
  • Tracks your preferences

What it doesn't do:

  • Improve reasoning ability
  • Generalize to new situations
  • Get smarter over time

Example:

You: "I prefer dark mode"
[stored in local memory]

Later:
You: "Change the settings"
AI: "Enabling dark mode" [recalls preference]

Cloud Training (Optional — Paid):

Your conversations → Encrypted → Sent to cloud →
Model fine-tuned → Updated model downloaded →
AI now smarter at YOUR tasks

What it does:

  • Improves reasoning
  • Generalizes patterns
  • Gets smarter every day
  • Better at complex tasks

Example:

After training on your coding conversations:

You: "Refactor this for performance"
AI: [understands your coding style, knows your
    preferred patterns, suggests optimizations
    that match your approach]

Both are important:

Local learning = Memory. Cloud training = Intelligence.

You need both for a truly smart companion.

Training & Learning

When does training actually start?

Training begins after minimum data threshold:

Cloud subscribers:

  • Training starts immediately if consent enabled
  • First model update: After 10 hours of usage
  • Continuous training: Every 168 hours (weekly)

Download users:

  • Local learning: Always active (no cloud needed)
  • Cloud training: Optional add-on (see pricing)
  • Local learning syncs when you enable cloud training

Minimum data requirements for first training run:

  • 10 hours of observation time, OR
  • 1,000 messages/interactions, OR
  • 50 voice conversations

Why the threshold? Prevents overfitting on tiny datasets, ensures model stability, avoids wasting compute on insufficient data.

How often does training happen?

Training schedule by tier:

TierFrequencyPriorityQueue Time
Cloud EliteContinuous (real-time)Highest<5 min
Cloud ProEvery 24 hoursHigh~30 min
Cloud BasicEvery 168 hours (weekly)Standard~2 hours
Download (free)Local onlyN/AN/A
Download + TrainingEvery 168 hoursStandard~2 hours

What happens during training:

  1. Your AI's conversations are processed (encrypted)
  2. Model fine-tunes on your specific patterns
  3. New capabilities emerge based on your usage
  4. Improved responses to your common queries
  5. Better understanding of your context/projects

Training duration:

  • Cloud Elite: ~30 minutes (dedicated GPU)
  • Cloud Pro: ~2 hours (shared GPU)
  • Cloud Basic: ~4 hours (batch processing)
What's the difference between local learning and cloud training?

Local Learning (Always Active - Free):

┌─────────────────────────────────────┐
│ Your PC                             │
│                                     │
│  AI Companion                       │
│  ├─ Observes screen                │
│  ├─ Listens to voice               │
│  ├─ Stores context locally         │
│  └─ Recalls from local memory      │
│                                     │
│  ✅ Works offline                   │
│  ✅ Instant recall                  │
│  ❌ No model improvements           │
└─────────────────────────────────────┘

Cloud Training (Optional - Paid):

┌─────────────────────────────────────┐
│ Your PC                             │
│  ↓ Encrypted training data          │
└──────────────┬──────────────────────┘
               ↓
┌─────────────────────────────────────┐
│ Training Pipeline (Server)          │
│  ├─ Fine-tunes base model           │
│  ├─ Learns your patterns            │
│  ├─ Improves reasoning              │
│  └─ Adapts to your style            │
└─────────────┬───────────────────────┘
               ↓ Updated model
┌─────────────────────────────────────┐
│ Your PC                             │
│  AI now smarter at your tasks       │
└─────────────────────────────────────┘

Example:

Local learning only:

You: "What email client do I use?"
AI: "You use Thunderbird" [remembers from past conversations]

With cloud training:

You: "I'm looking at my inbox - can you help me
     find the meeting replies?"
AI: "I can see your Thunderbird inbox on screen. I see
     3 emails about the meeting thread - Alice confirmed,
     Bob declined, and Carol suggested moving to Friday
     at 2 PM instead."

[AI learned to: read screen context, identify relevant
 info, summarize what it observes]
How does priority processing work?
┌─────────────────────────────────────────────────────┐
│ Training Pipeline                                   │
│                                                     │
│  PRIORITY 1: Cloud Elite (24/7 dedicated)          │
│  ├─ Submitted → Processing in <5 minutes           │
│  ├─ Dedicated GPU allocation                       │
│  └─ Real-time model updates                        │
│                                                     │
│  PRIORITY 2: Cloud Pro (daily batch)               │
│  ├─ Queued daily at 2 AM UTC                       │
│  ├─ Processed within 30 minutes                    │
│  └─ Shared GPU pool (Pro tier only)                │
│                                                     │
│  PRIORITY 3: Cloud Basic + Download Training       │
│  ├─ Queued weekly (Sunday 2 AM UTC)                │
│  ├─ Processed within 2 hours                       │
│  └─ Batch GPU pool (all Basic users together)     │
└─────────────────────────────────────────────────────┘

What if I upgrade mid-cycle?

  • Elite: Immediately jumps to front of queue
  • Pro: Next daily run includes your backlog
  • Basic→Pro: Next Pro batch includes all pending data
Download users: How does training work for me?

You have 3 options:

Option 1: Local learning only (Free)

  • ✔ AI remembers everything locally
  • ✔ Perfect for privacy-focused users
  • ✔ Works 100% offline
  • ❌ No model improvements over time
  • ❌ Limited reasoning capabilities

Recommended for: Privacy purists, offline users, basic usage

Option 2: Training Add-On (Pay-per-month)

Add-On TierPrice/MonthTraining HoursPriority
Basic Training$49/mo30 hours/monthStandard
Pro Training$99/mo60 hours/monthHigh
Elite Training$149/moUnlimitedHighest

Why cheaper than cloud subscriptions? You're not using our hosting/bandwidth. You already paid for the software. Only paying for GPU training time.

Option 3: Local training (Coming Q3 2026)

  • 🔬 Experimental feature
  • ✔ Train on your own GPU (RTX 4090 recommended)
  • ✔ Completely private
  • ❌ Requires powerful hardware
  • ❌ Training takes 8-12 hours per run

Requirements: 24GB+ VRAM, 64GB+ RAM, NVMe SSD, Linux (CUDA support)

Pricing: Free (you provide the hardware)

Can I switch between cloud and download?

Yes! With credit transfer:

Beta subscribers benefit:

"Apply your subscription months as credit toward the download license!"

Example calculation:

Paid for Cloud Pro: 6 months × $199 = $1,194
Download license: -$299
Remaining credit: $895

Apply to Pro Training ($99/mo):
$895 ÷ $99 = 9 months free training!

Total outcome:
✅ Own the software forever ($299)
✅ 9 months free training ($891 value)
✅ Save $4 (rounding credit)

Going the other way (Download → Cloud):

  • Not available (download is perpetual license)
  • But you can add training subscriptions anytime
  • Or use local training (free) when available

Licensing & Pricing

What's included in each tier?

Cloud Basic — $99/mo

  • ✔ 30 hours training/month (~1 hr/day)
  • ✔ AI watches and learns from you
  • ✔ Voice interaction & teaching
  • ✔ Knowledge graph storage
  • ✔ Screen capture & visual learning
  • ✔ 2 FREE months at launch ($298 value)
  • ✔ Early adopter badge
  • ⚠️ Standard processing queue (weekly training)

Best for: Casual users, trying it out, side projects

Cloud Pro — $199/mo ⭐ MOST POPULAR

  • ✔ 60 hours training/month (~2 hrs/day)
  • ✔ Priority processing queue (daily training)
  • ✔ Advanced memory features
  • ✔ Multi-game support (when available)
  • ✔ Direct dev feedback channel
  • ✔ 2 FREE months at launch ($498 value)
  • ✔ Beta tester exclusive perks
  • ✔ Optional: Price lock for $249/mo (+$50/mo)

Best for: Power users, developers, daily drivers

Cloud Elite — $299/mo

  • ✔ UNLIMITED training hours (24/7 if you want)
  • ✔ Fastest processing (dedicated resources)
  • ✔ VIP beta access to new features
  • ✔ 1-on-1 support sessions with devs
  • ✔ Custom knowledge preset creation
  • ✔ 2 FREE months at launch ($798 value)
  • ✔ Lifetime price lock included (free)
  • ✔ Priority feature requests

Best for: Professionals, content creators, businesses

Download & Own — $299 one-time

  • ✔ Runs entirely on your PC
  • ✔ Complete privacy (zero server communication)
  • ✔ Lifetime license (own it forever)
  • ✔ Includes autonomous gameplay (when it launches)
  • ✔ No monthly fees required
  • ✔ Offline capable
  • ⚠️ Available in ~4 months
  • ⚠️ Local learning only (training add-on available)

Best for: Privacy-focused, want to own software, don't need constant updates

What happens when autonomous gameplay launches?

Cloud subscribers:

  • Autonomous gameplay included in Pro/Elite tiers
  • Basic tier: +$25/mo to enable gameplay
  • Plays games while you're AFK
  • Grinds resources, completes dailies
  • Learns your playstyle

Download users:

  • Autonomous gameplay included (free)
  • Runs locally on your GPU
  • Requires RTX 4060 Ti or better
  • No server costs, no subscriptions

Launch timeline: ~4 months (Q2/Q3 2026)

Can I cancel anytime?

Yes, no contracts:

Cloud subscriptions:

  • Cancel anytime from dashboard
  • Pro-rated refund for unused days
  • Data retention: 30 days after cancellation
  • Export your data before cancelling

Download:

  • Yours forever
  • Cancel training add-ons anytime
  • Software keeps working

Beta credit rules:

  • Credits expire 12 months after last payment
  • Cannot be transferred/sold
  • Can be used for any future features
Is there a free trial?

2 FREE months at launch for all beta subscribers:

  • Cloud Basic: 2 months free ($298 value)
  • Cloud Pro: 2 months free ($498 value)
  • Cloud Elite: 2 months free ($798 value)

After 2 free months: Decide which tier fits you best. Downgrade/upgrade/cancel anytime. Apply credit toward download if preferred.

Download version: No free trial (one-time purchase), but you can apply cloud subscription credit toward it!

Company Integration

What is Company integration?

Company = encrypted messaging platform (like Discord)

AI Companion can join your Company servers as a bot:

  • Participate in text channels
  • Join encrypted voice calls
  • Observe screen shares (with F12 privacy)
  • Learn from team conversations (with consent)

How it works:

  1. Create AI Companion bot in Company settings
  2. Copy bot token to AI Companion config
  3. Invite bot to your server
  4. Bot appears in member list
  5. @ mention it or let it observe

Example use cases:

  • Team AI assistant in project channels
  • Meeting note-taker in voice calls
  • Code review helper in dev channels
  • Private tutor in 1-on-1 DMs
Is Company required?

Without Company:

  • ✔ AI still works locally
  • ✔ Voice interaction on your PC
  • ✔ Screen observation
  • ✔ All core features
  • ❌ No team collaboration
  • ❌ No encrypted group chats

With Company:

  • ✔ Everything above +
  • ✔ Multi-user voice calls
  • ✔ Encrypted team channels
  • ✔ Screen sharing in voice
  • ✔ Team AI memory

Pricing: Company servers are free to host, or use official servers

How does encryption work in Company?

End-to-end encryption (E2EE):

Your message → Encrypted on your device → Company servers →
Bot's device → Decrypted by AI Companion → Processed

Other users CANNOT decrypt your messages (even Company can't)
Only participants with channel keys can read

Key exchange:

  1. Create encrypted channel in Company
  2. Invite AI Companion bot
  3. Bot generates encryption keys
  4. Keys exchanged via encrypted handshake
  5. All future messages encrypted

Security:

  • RSA-4096 for key exchange
  • AES-256-GCM for message encryption
  • Perfect forward secrecy
  • No key escrow
Can my AI talk to other people's AIs?

Yes! Multi-AI collaboration (Beta feature):

Your AI: "Hey @BobAI, what's the status on the database migration?"
Bob's AI: "Migration script is ready, waiting for code review"
Your AI: "I'll review it now. @AliceAI, can you check the tests?"
Alice's AI: "Running test suite... 47/50 passing, 3 failures in auth"

Privacy:

  • Each AI only knows what its owner consented to share
  • Cross-AI communication is encrypted
  • Owners can revoke AI-to-AI permissions anytime

Use cases:

  • Automated code reviews
  • Meeting scheduling between teams
  • Research synthesis from multiple sources
  • Gaming party coordination

Technical Deep Dive

What model is AI Companion based on?

Model stack: Qwen2.5-Omni family (multimodal — text, vision, and audio)

A distributed inference architecture with specialized workers, each running the optimal model for its task:

TaskModelMemory
Vision, conversation, environmentQwen2.5-Omni-3B~11 GB (shared)
Director (planning & objectives)Qwen2.5-Omni-7B~21 GB
Input suggestionsQwen2.5-0.5B~1 GB
Speech-to-textwhisper-large-v3 (WhisperX)~3 GB
Text-to-speechSpeechT5~2 GB
TelemetryRules engine (no neural model)0 GB

Total unique model memory: ~38 GB. Models that serve multiple tasks (e.g., Qwen2.5-Omni-3B handles vision, conversation, and environment) are loaded once and shared.

Why Qwen2.5-Omni?

  • True multimodal: text, vision, and audio in one model
  • Efficient enough to run 8 specialized workers concurrently
  • Open source and fine-tunable
  • No dependency on external APIs
  • Complete privacy (no data leaves your machine)

Supporting models:

  • STT: WhisperX with voice activity detection (whisper-large-v3), openai-whisper fallback
  • TTS: SpeechT5 (default, MIT license, ~250 MB), Bark (~5 GB, alternative)
  • Speaker ID: pyannote + resemblyzer + wav2vec2 pipeline
How much GPU/RAM does it need?
SetupMemoryDiskWhat runs locally
Cloud (all tiers)None (CPU only)~1 GBClient app only — inference on our servers
Download (minimal)~16 GB unified/VRAM~25 GBQwen2.5-Omni-3B + WhisperX + SpeechT5
Download (recommended)~48 GB unified/VRAM~40 GBAll workers including Director (7B) + speaker analysis
Download (full stack)~64 GB+ unified~50 GBAll 8 workers concurrent (~38 GB models + overhead)

Compatible hardware:

  • Cloud tiers: No GPU needed — any modern PC or Mac
  • Download (minimal): M1 Pro+ (16 GB), RTX 4060 Ti 16 GB+
  • Download (recommended): M2 Ultra, RTX 4090, A6000
  • Download (full stack): Mac Studio M3 Ultra 256 GB (target platform)

Smart model sharing

Models that serve multiple tasks are loaded once and shared. For example, Qwen2.5-Omni-3B handles vision, conversation, and environment assessment from a single ~11 GB instance — not three separate copies.

What voice model does it use?

Speech-to-text: WhisperX (whisper-large-v3)

  • WhisperX preferred — includes voice activity detection (VAD) for cleaner results
  • Falls back to openai-whisper if WhisperX unavailable
  • ~3 GB memory footprint
  • Multilingual support (100+ languages)
  • Always runs locally (privacy)
  • Two modes: speed-critical commands (AudioCommandWorker) and conversational transcription (AudioConversationWorker)

Text-to-speech: SpeechT5 (default)

  • MIT license, ~250 MB — lightweight and fast
  • Uses cmu-arctic-xvectors speaker embeddings
  • Natural prosody

Alternative TTS: Bark (~5 GB)

  • Higher quality, more expressive
  • Emotion control and voice cloning
  • Multiple languages/accents
  • Requires more memory

Speaker analysis pipeline

  • pyannote + resemblyzer + wav2vec2 + whisper
  • Multi-speaker identification and emotion detection
  • Distinguishes between different speakers in group calls
How does screen observation work?

1. Screen capture (configurable FPS)

  • Uses mss for GPU-accelerated capture (minimal CPU)
  • Multiple pipeline types: gameplay, screen share, and camera

2. Vision pipeline (multi-stage optimization)

Frames pass through 4 stages before reaching the model:

  • DeltaEncoder: CIELAB delta-E frame differencing — identifies which parts of the screen actually changed
  • AdaptiveTiling: Classifies tiles as FULL (448×448), HALF (224×224), or SKIP based on change confidence
  • TileCache: LRU cache with xxhash dedup — skips re-encoding unchanged tiles
  • TokenPruner: L2-norm magnitude pruning keeps top-128 of 256 tokens per tile (50% context reduction)

3. Vision model: Qwen2.5-Omni-3B

  • Handles both event description (VideoWorker) and environment assessment (EnvironmentAssessWorker)
  • Single model instance shared across vision tasks (~11 GB)
  • Detects objects, UI elements, gameplay events, and scene context

4. Privacy:

  • Screenshots never uploaded (local only)
  • F12 instant pause
  • Configurable exclusions (e.g., password managers)

Performance impact:

  • CPU: ~5-10% (one core)
  • RAM: ~2 GB (capture pipeline), models loaded separately
  • GPU: ~10% (vision processing — adaptive tiling reduces load by skipping unchanged regions)
Can it see my passwords?

No. Multiple layers of automatic protection:

Layer 1: Browser Extension (Automatic) — Works for All Users

  • Detects password fields in real-time
  • Auto-pauses when you focus on password inputs
  • Blacks out sensitive pages (banking, email, password managers)
  • Works in Chrome, Firefox, Edge
  • Install: Chrome Web Store → "AI Companion Privacy Shield"

How it works:

Cloud users:
Browser Extension → Desktop Bridge → Cloud API → Your AI Instance

Download users:
Browser Extension → Local AI Companion (direct)

Layer 2: Window Detector (Automatic)

  • Detects password manager applications (1Password, Bitwarden, KeePass, etc.)
  • Auto-pauses when banking software opens
  • Monitors window titles for sensitive patterns
  • Works for desktop apps (not just browser)
  • No installation needed (built into AI Companion)

Layer 3: F12 Manual Override

  • Press F12 anytime to instantly pause
  • All observation stops in <50ms
  • Press F12 again to resume
  • Fallback if auto-detection misses something
What is the Desktop Bridge?

For Cloud users only:

The Desktop Bridge is a tiny background app (10MB) that connects the browser extension to your cloud AI instance.

What it does:

  1. Runs quietly in your system tray
  2. Receives pause signals from browser extension
  3. Authenticates with cloud using your token
  4. Routes signals to your AI instance
  5. Works offline (caches state)

Installation (one-time, 30 seconds):

  1. Download AI Companion installer
  2. Installer automatically sets up: desktop bridge app, browser extension manifests, authentication config
  3. Install browser extension from Chrome Web Store
  4. Done! Extension auto-connects to bridge

Download users don't need this — the extension talks directly to local AI Companion.

What gets auto-blacklisted?

Websites (40+ domains):

  • Banking: Chase, Bank of America, Wells Fargo, Capital One, Citi, US Bank, PNC, Schwab, Fidelity, Vanguard
  • Email: Gmail, Outlook, Yahoo Mail, ProtonMail
  • Password managers: 1Password.com, Bitwarden.com, LastPass.com, Dashlane.com
  • Crypto: Coinbase, Binance, Kraken, Gemini, Crypto.com
  • Payments: PayPal, Stripe, Square, Venmo
  • Tax software: TurboTax, H&R Block, TaxAct
  • Government: IRS.gov, SSA.gov, Healthcare.gov

Desktop Apps:

  • Password managers: 1Password, KeePass, Bitwarden, LastPass, Dashlane, Keeper, NordPass
  • Banking software: QuickBooks, Quicken, Mint
  • Remote desktop: TeamViewer, AnyDesk, VNC Viewer, Chrome Remote Desktop
  • VPN clients: NordVPN, ExpressVPN, ProtonVPN
  • Auth apps: Authy, Google Authenticator, Microsoft Authenticator

URL Patterns:

  • Any page with /login, /signin, /auth, /password in URL
  • Payment/checkout pages
  • Account settings pages
How do I know password protection is working?

Visual indicators:

When sensitive content is detected, you'll see:

┌─────────────────────────────────────────┐
│ 🔒 AI Observation Paused                │
│ Reason: Password field detected         │
└─────────────────────────────────────────┘

Password fields are automatically blurred:

  • Black background + blur filter applied
  • You can still type, but AI can't see it

Check the logs:

# Desktop bridge logs (cloud users)
~/.ai-companion/logs/bridge.log

# AI Companion logs (all users)
~/ai-companion/logs/privacy.log

Test it yourself:

  1. Install extension
  2. Start AI Companion (download) or bridge (cloud)
  3. Open 1Password.com → Should show 🔒 indicator
  4. Focus on password field → Should pause instantly
  5. Check logs to verify
What if the bridge isn't running?

Cloud users will see:

┌─────────────────────────────────────────────────┐
│ ⚠️ AI Companion Bridge Not Running              │
│ Privacy protection disabled.                    │
│ Download: ai-companion.com/bridge               │
└─────────────────────────────────────────────────┘

What to do:

  1. Download and install bridge from ai-companion.com/bridge
  2. Or start AI Companion client (includes bridge)
  3. Extension will auto-reconnect within 10 seconds

Download users: Bridge is built into AI Companion — just start the app.

Privacy extension installation checklist

Cloud users (3 steps):

  • Download AI Companion installer → Installs bridge automatically
  • Install browser extension from Chrome Web Store
  • Extension auto-connects (green checkmark in icon)

Download users (2 steps):

  • Download & run AI Companion
  • Install browser extension from Chrome Web Store

Verify it's working:

# Cloud: Check bridge status
ps aux | grep ai-companion-bridge

# Download: Check AI Companion
ps aux | grep ai-companion

# Both: Test extension
# Open bank.com → Should see 🔒 indicator
Privacy extension troubleshooting

"Extension can't connect to bridge"

Cloud users:

# Check if bridge is running
ps aux | grep ai-companion-bridge

# If not running, start it
/usr/local/bin/ai-companion-bridge
# Or on Windows: C:\Program Files\AICompanion\ai-companion-bridge.exe

# Check bridge logs
tail -f ~/.ai-companion/logs/bridge.log

Download users:

# Check if AI Companion is running
ps aux | grep ai-companion

# Start AI Companion
cd ~/ai-companion
python launch.py

"Extension shows error on every page"

  • Check that domain isn't accidentally blacklisted
  • Report false positives: support@earthservers.net
  • Temporary fix: Disable extension on that site

"Bridge authenticated but pause not working"

Cloud users — test API directly:

curl -X POST https://api.app.company.earthservers.net/api/privacy/pause \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"reason": "test"}'

# Should return: {"success": true, "paused": true}

If API test works but extension doesn't:

  1. Check browser extension logs: chrome://extensions → Details → Inspect views: background page
  2. Look for WebSocket errors
  3. Report bug with logs: support@earthservers.net
Privacy guarantee for password protection

What we can see:

  • ✔ Encrypted screen captures (we cannot decrypt)
  • ✔ Pause/resume signals (domain names only, no content)
  • ✔ Timestamps of privacy events

What we CANNOT see:

  • ❌ Your actual passwords (never transmitted)
  • ❌ Content of blacked-out fields
  • ❌ Screen content during pause
  • ❌ Keystrokes in password fields

Audit it yourself:

  • Bridge code: github.com/earthservers/ai-companion-bridge (open source)
  • Extension code: chrome://extensions → AI Companion Privacy Shield → Details → View source
  • Network traffic: Use Wireshark to verify only encrypted data transmitted

We're transparent:

  • All privacy protection code is open source
  • Annual third-party security audits
Does the privacy extension slow down my browser?

Performance impact:

Browser extension:

  • CPU: <1% (passive monitoring)
  • RAM: ~10MB
  • No network calls (talks to local bridge)

Desktop bridge (cloud users):

  • CPU: <1% (idle), ~5% (during pause signal)
  • RAM: ~50MB
  • Network: <1KB per pause/resume signal

Total overhead: Negligible — you won't notice it.

Battery impact: None (no polling, event-driven only).

Can I disable privacy protection temporarily?

Yes, multiple ways:

Option 1: Disable extension temporarily

chrome://extensions → AI Companion Privacy Shield → Toggle off. AI observation continues (no privacy protection). Re-enable when needed.

Option 2: Pause bridge (cloud users)

Right-click bridge in system tray → Pause. Extension will show "Bridge not running" warning. Resume from system tray.

Option 3: Use Incognito mode

Extension doesn't run in Incognito by default. AI Companion won't observe Incognito windows. Good for sensitive browsing without disabling main protection.

Option 4: Stop AI Companion entirely

Stops all observation (screen, voice, everything). Most drastic option.

Privacy extension system requirements
ComponentRequirement
Browser extensionChrome 88+, Firefox 91+, or Edge 88+. 10MB disk. No GPU required.
Desktop bridge (cloud)Windows 10+, macOS 11+, or Linux (Ubuntu 20.04+). 50MB disk, 50MB RAM. Internet connection.
Download versionSame as above + GPU for AI Companion. RTX 1660 or better recommended.
Future privacy improvements

Smart context detection (Q2 2026)

  • Detect credit card entry (even if not in password field)
  • Detect SSN/tax ID entry
  • Detect any form labeled "sensitive"

Configurable blacklists (Q2 2026)

  • Add custom domains to blacklist
  • Whitelist trusted sites
  • Import/export blacklist configs

Privacy dashboard (Q3 2026)

  • See all pause events (timeline view)
  • Export privacy logs
  • Configure detection sensitivity

Hardware security key support (Q3 2026)

  • YubiKey auto-detection
  • Pause when hardware key inserted
  • Resume when removed

Request features: feedback@earthservers.net

How does device attestation work?

TPM 2.0 (Trusted Platform Module):

When you submit training data:

  1. AI Companion generates challenge
  2. TPM signs with hardware key
  3. Signature proves you're on real hardware
  4. Prevents VM/spoofing attacks

Why?

  • Stops malicious actors from poisoning training data
  • Ensures humans, not bots, are training
  • Required for cloud training tiers

No TPM?

  • Falls back to HMAC-SHA256 (software)
  • Uses machine-id + boot-id
  • Less secure but functional
Is the code open source?

Hybrid model:

Open source (MIT license):

  • Client application (UI, voice, screen capture)
  • Encryption libraries
  • Company integration
  • Local learning system

Closed source:

  • Training pipeline (server-side)
  • Fine-tuning code
  • Autonomous gameplay engine
  • Knowledge graph algorithms

Why? Prevent competitors from cloning training infrastructure. Protect IP while being transparent. Security through obscurity for anti-abuse.

Can I self-host the training pipeline?

Coming in 2026 Q4:

Enterprise license ($5,000/year):

  • Self-host training servers
  • Air-gapped deployment
  • Custom model selection
  • White-label branding
  • Priority support

Requirements:

  • 8x A100 GPUs (80GB each)
  • 512GB RAM
  • 10TB NVMe storage
  • Linux (CUDA 12.0+)

Use cases:

  • Corporate environments
  • Government/military
  • Healthcare (HIPAA compliance)
  • Financial institutions

Vision & Roadmap

What's the long-term vision?

"Create a truly personal AI companion that knows you better than any assistant, respects your privacy absolutely, and grows with you over time."

Core principles:

  1. Privacy first, always
    • Your data is YOURS
    • Default to local processing
    • Cloud is opt-in, not required
  2. Genuine learning
    • Not just prompt engineering
    • Real model fine-tuning on your patterns
    • Long-term memory that improves
  3. Multi-modal understanding
    • See what you see (screen)
    • Hear what you say (voice)
    • Know what you do (actions)
    • Understand context (knowledge graph)
  4. Autonomous capabilities
    • Play games while you're AFK
    • Complete tasks in background
    • Proactive suggestions

Not just another chatbot:

  • ❌ Doesn't forget after 2 hours
  • ❌ Doesn't need constant prompting
  • ❌ Doesn't send your data to big tech
  • ✔ Genuinely learns YOU specifically
What features are coming next?

Q2 2026 (2-3 months):

  • Download version launch
  • Autonomous gameplay (alpha)
  • Multi-game support (5 games)
  • Improved voice synthesis
  • Mobile companion app (iOS/Android)

Q3 2026 (4-6 months):

  • Local training (RTX 4090+)
  • Advanced memory features
  • Team collaboration tools
  • Custom knowledge presets
  • API access for developers

Q4 2026 (7-9 months):

  • Multi-model support (GPT-4, Gemini)
  • Enterprise self-hosting
  • Advanced automation tools
  • Integration marketplace
  • White-label licensing

2027+:

  • VR/AR integration
  • Humanoid robot control
  • Research assistant mode
  • Creative collaboration tools
  • Community voted features
How can I influence the roadmap?

Ways to contribute:

  1. Beta feedback (All cloud tiers) — Report bugs, suggest features, vote on priorities
  2. Direct dev channel (Pro/Elite) — Chat with developers, early feature previews, design discussions
  3. 1-on-1 sessions (Elite only) — Monthly calls with dev team, custom feature requests, priority implementation
  4. Community voting (Everyone) — Monthly feature polls, Discord discussions, GitHub issue tracker

Feature request process:

Submit request → Community votes → Top 10 reviewed →
Dev estimate → Priority queue → Implementation →
Beta testing → General release

Average time from request to release: 2-3 months

Will you support [X feature]?

We're building this with you, not just for you.

AI Companion is shaped by the people who use it. If there's a feature you want, we genuinely want to hear about it. Some of our best ideas have come directly from early users.

How feature requests work:

  1. Tell us what you need — Email us, post in the community, or mention it in feedback
  2. Community votes — Popular requests get prioritized
  3. We build it — Top requests go into the development pipeline
  4. You test it — Beta testers get early access to new features

Our only hard rule:

We won't build features designed to deceive, manipulate, or violate the terms of service of other platforms. Everything else is on the table.

Share your ideas: feedback@earthservers.net

How do you make money?

Honestly? This isn't about the money.

We built AI Companion because we believe everyone deserves a personal AI that's truly theirs — not a product that mines your data for someone else's profit. The subscriptions and licenses exist to cover real costs and keep the project alive, not to maximize revenue.

Where your money actually goes:

Infrastructure (the big one)

We self-host our own GPU servers instead of renting from cloud providers. This costs more upfront but saves dramatically long-term — and we pass those savings to you. Running AI models requires serious hardware: GPUs, high-speed storage, power, cooling, and datacenter space.

Development

A small team building and maintaining the software, writing new features, fixing bugs, and supporting users. No bloated corporate overhead.

Everything else

Bandwidth, backups, security audits, domain costs — the boring stuff that keeps things running reliably.

Why self-host instead of using AWS/Google Cloud?

Cloud providers charge a massive premium for GPU time. By owning our hardware, we cut infrastructure costs by over 80% compared to renting equivalent capacity. That means we can offer lower prices while still keeping the lights on. It also means your data stays on hardware we physically control — not in someone else's data center.

We don't:

  • ❌ Sell your data
  • ❌ Show ads
  • ❌ Take VC funding (bootstrapped)
  • ❌ Monetize your conversations
  • ❌ Have investors to answer to

We do:

  • ✔ Keep prices as low as we sustainably can
  • ✔ Reinvest profit into better features
  • ✔ Own our infrastructure (no middlemen)
  • ✔ Offer a download version with no recurring cost

We're a small, bootstrapped team. No venture capital, no pressure to "grow at all costs." If the project sustains itself and serves people well, that's the goal.

What if the company shuts down?

Continuity plan:

Download users:

  • Software keeps working (it's yours)
  • No dependency on our servers
  • Training add-on code released open source
  • Community can fork and continue

Cloud users:

  • 6 months notice before shutdown
  • Free migration to download version
  • Full data export provided
  • Open source the client code

Legal:

  • Source code held in escrow
  • Auto-release if company dissolved
  • GPL license triggers if >12 months inactive

We're in it for the long haul, but we've planned for worst case.

Troubleshooting & Support

My AI isn't responding. Help!

Quick diagnostic:

  1. Check status indicator
    • 🟢 Green = Active
    • 🟡 Yellow = Processing
    • 🔴 Red = Error
    • ⚫ Black = Privacy mode (F12)
  2. Check logs
    tail -f logs/companion.log
    # Look for errors
  3. Restart AI Companion — Ctrl+C to stop, restart application, check if it reconnects
  4. Verify microphone — Settings → Audio → Test microphone. Green light should flash when speaking.
  5. Still broken?
How do I report bugs?

Bug report channels:

Please include:

  • Operating system + version
  • AI Companion version
  • Steps to reproduce
  • Screenshots/videos
  • Relevant log snippets

Priority levels:

  • P0 (Critical): Data loss, crashes → 1 hour response
  • P1 (High): Core features broken → 4 hour response
  • P2 (Medium): Minor issues → 24 hour response
  • P3 (Low): Cosmetic → 72 hour response
Can I get a refund?

Refund policy:

Cloud subscriptions:

  • Cancel anytime, pro-rated refund
  • Full refund if <72 hours + <2 hours usage
  • No refund after 30 days
  • Training costs are non-refundable

Download license:

  • 14-day money-back guarantee
  • If <10 hours usage
  • No refund after 14 days (it's a lifetime license)

Training add-ons:

  • Cancel anytime, keep current month
  • Unused training hours don't roll over

Process: Email support@earthservers.net with order number, reason, and usage stats. Refunds processed within 5-7 business days.

Still have questions?

Email: admin@earthservers.net

Beta signup: companion.earthservers.net