ii.inc (Intelligent Internet) - Sovereign AI Infrastructure Platform
Company Overview
ii.inc (Intelligent Internet) is an open-source, decentralized AI infrastructure platform founded by Emad Mostaque. The company is building sovereign AI governance frameworks and autonomous reasoning agents designed to democratize access to advanced AI while maintaining user control, transparency, and data sovereignty.
Mission
“The best open intelligence for everyone” - ii.inc emphasizes building transparent, auditable AI systems that empower users rather than concentrate control in centralized organizations.
Core Philosophy
- User Sovereignty: Users control their AI agents via cryptographic keys
- Open-Source Transparency: All core functionality is auditable and transparent
- Decentralization: Distributed infrastructure over centralized control
- Alignment First: Safety and governance built into the foundation
Product Ecosystem
Tier 1: Core Agents
II-Agent V1 (Production-Ready, Feb 2026)
The flagship autonomous reasoning agent for real-world work across multiple domains.
Performance Metrics:
- Terminal Bench: 61.8% (coding/terminal tasks)
- SWE Bench Pro: 45.1% (software engineering)
- GAIA Validation: 73% (general AI assistant)
- FRAMES Test: 80% (research)
Core Capabilities:
- Multi-model support (Claude, Gemini, GPT-5 in single conversation)
- Code generation and execution
- Full-stack web development
- Data analysis and visualization
- Research and fact-checking
- Browser automation (Playwright-powered)
- Document skills (PDF, Excel, Word, PowerPoint)
- Image generation (Nano Banana Pro, GPT Image 1.5, Imagen 4.0)
Architecture:
- Partner Agent: Runs on user’s edge device, interprets goals, maintains communication, drafts plans
- Principal Agents: Orchestrate execution teams for specific project goals
- Associate Agents: Short-lived workers executing discrete tasks
- All agents remain fully open-source (can call proprietary endpoints when needed)
Key Features:
- BYOK (Bring Your Own Keys): Users maintain full control over API costs and model selection
- Plan Mode: Visualize and modify project plans before execution
- Universal Connectors: GitHub, Slack, Gmail, Google Workspace, Notion, Discord, Dropbox, Canva
- Research Integration: Tight integration with II-Researcher for multi-step investigations
- Code Execution: Full terminal access for data processing and analysis
II-Researcher
Autonomous research agent framework for complex, multi-step inquiries.
Capabilities:
- Real-time literature review
- Hypothesis generation
- Data visualization
- Source triangulation
- Structured report generation
Performance:
- GAIA Validation Subset: 73%
- FRAMES Test Set: 80%
II-Medical (Specialized)
Compact, open-source 8-billion parameter model for clinical reasoning and medical decision support.
Characteristics:
- Safety-optimized deployment
- Zero inference costs
- Domain-specialized training
- Average accuracy 70.4% across 10 medical benchmarks
- Suitable for healthcare applications requiring domain expertise
Tier 2: Knowledge & Infrastructure
II-Commons
Context suite for human-AI collaboration featuring the II-Chronicle engine.
Features:
- Auto-timeline generation from unstructured conversations
- Conflict resolution in multi-agent environments
- Shared workspaces for collaboration
- II-Chronicle Engine: Converts research questions into source-grounded timelines
- 92% event-alignment accuracy
- Over 3.2 million embedded Arxiv abstracts
- Scientific research integration
Common Ground Protocol
Framework for structured human-AI collaboration emphasizing partnership over replacement.
Architecture:
- Three-layer professional team (mirroring elite consulting models)
- Declarative agent personas via YAML profiles
- Multi-view observability (Flow, Kanban, Timeline views)
- Persistent procedural memory (local audit trail, learning, reuse)
- Transparent multi-agent workflows
Focus: Augmenting human judgment rather than replacing it
Tier 3: Specialized Models
II-Thought
Reasoning and reinforcement learning model with extensive dataset.
Dataset: 340,000 samples covering:
- Mathematics
- Code
- Science
- Commonsense reasoning
Performance:
- AMC23: 73.8% accuracy
- Math500: 85.8% accuracy
II-Explorer (Implied from ecosystem)
Data exploration and analysis tools integrated into the platform
Governance Architecture
Guardian Lattice
Advanced II-Agents operating in three specialized oversight roles:
Sentinels:
- Stream runtime logs
- Meter energy attestations
- Replay proof-of-burn receipts
- Monitor system activity
Advisers:
- Digest Sentinel telemetry
- Publish human-readable risk reports
- Provide safety analysis
Implementers:
- Hold time-locked pause key
- Can stop block production in emergencies
- Execute emergency protocols
Rotation: Automatic every six months via stake-weighted elections
Oracle Council
Fifteen-seat governing body responsible for parameter tuning during Phase Three.
Responsibilities:
- Benefit class definitions
- SLA threshold setting
- Job-class registry management
Process:
- Proposals require minimum 10 Council votes
- 30-day grace period before activation
- Decisions executed via decomposed parallel tasks
Alignment System
Status Score integrates:
- Proof-of-personhood status
- Compute contributions
- External licenses
- Personal learning preferences
Personalization: Each citizen controls one II-Agent that learns:
- Reading level preferences
- Accessibility needs
- Ethical preferences
- Domain-specific contexts
Core safety rules remain identical across all tiers; optional upgrades are fully optional and self-attestation based.
Technical Stack & Integrations
Multi-Model Support
Users can work across:
- Claude Sonnet / Opus 4.5
- Gemini 3 Pro
- GPT-5.2
- Custom models via API keys (BYOK)
- Open-source models
Full context retention in single conversation thread.
Integration Ecosystem
Universal Connectors:
- GitHub (code repositories, PRs, issues)
- Slack (team communication)
- Gmail (email automation)
- Google Workspace (docs, sheets, slides)
- Notion (knowledge management)
- Discord (community channels)
- Dropbox (file storage)
- Canva (design collaboration)
Execution Capabilities:
- Playwright browser automation
- Shell command execution
- Python/Node.js code execution
- GPU training runs
- Data processing pipelines
Research Integration
II-Researcher integration enables:
- Fast Research: Quick answers with source validation
- Deep Research: Multi-step investigations with triangulation
- Hypothesis generation
- Report generation with source citations
Sovereignty Architecture
User Control Model
II-Accounts:
- Single account per user via WebAuthn key
- Key hash anchored on ledger
- Cryptographic ownership
- No KYC required for core functionality
Zero KYC Design:
- Core functionality available without identity verification
- Optional proofs (personhood, external APIs) grant incremental privileges
- Privacy-first approach
API Key Management:
- API keys remain on user’s device
- Never broadcast to network
- User maintains complete control over costs and service selection
- Supports specialized domain models (legal, medical, custom)
Open-Source Foundation
All core components remain fully open-source:
- Agent code is auditable and reproducible
- Security can be independently verified
- No hidden functionality or telemetry
- Users can fork, modify, deploy independently
Economic Model
Foundation Coin
The platform operates under a new economic model for the intelligence age:
- Rewards contributions to the network
- Aligns incentives across participants
- Supports sustainable development
Cost Structure
BYOK Pricing: Users pay only for models they choose
- GPT-5.2: Market rates (~$3-15 per 1M tokens)
- Claude: Market rates (~$3-15 per 1M tokens)
- Custom models: Varies by provider
- Open-source local models: Free
Compute Credits: Via Foundation Coin rewards for network contributions
SAGE (Sovereign AI Governance Engine)
ii.inc has announced SAGE as a framework for policy in the intelligence age:
- Governance policies for autonomous AI systems
- Regulatory compliance mechanisms
- Decentralized decision-making protocols
- Intelligence oversight and accountability
Competitive Positioning
vs Traditional AI (OpenAI, Anthropic, Google)
✅ ii.inc Advantages:
- Open-source (auditable)
- User-sovereign (no central control)
- Specialized models (medical, domain-specific)
- Governance-first design
- No vendor lock-in
❌ Limitations:
- Younger platform
- Smaller community than closed-source alternatives
- Less marketing/brand recognition
vs Other Open-Source Agents (Agent Zero, LangChain)
✅ ii.inc Advantages:
- Governance architecture (Guardian Lattice, Oracle Council)
- Specialized domain models (medical, researcher)
- User sovereignty (cryptographic keys)
- Alignment-first design
- Better organization and funding (founded by Emad Mostaque)
❌ Limitations:
- More complex governance (steep learning curve)
- Less established in developer community
- Newer platform
vs Decentralized AI (Others)
✅ ii.inc Advantages:
- Clear product roadmap
- Production-ready agents
- Governance frameworks
- Founder credibility (Emad Mostaque)
Use Cases
For Individuals
- Sovereign personal AI assistant (fully under their control)
- Research automation with source verification
- Content creation and analysis
- Code generation and development
- Data analysis and visualization
For Organizations
- Domain-specific AI (medical, legal, engineering)
- Compliant AI deployment (no data to third parties)
- Decentralized team collaboration via Common Ground
- Custom agent teams for complex workflows
For Researchers
- Open-source reasoning models and datasets
- Autonomous research workflows
- Hypothesis generation and testing
- Benchmarking against public datasets
For Regulators/Governments
- Governance frameworks via SAGE
- Transparent AI systems (auditable)
- Decentralized oversight
- Policy implementation mechanisms
Development Roadmap (Inferred)
Phase 1-2 (Current):
- II-Agent V1 production release
- II-Commons and II-Researcher integration
- BYOK and multi-model support
- Universal connectors expansion
Phase 3:
- Oracle Council governance activation
- Enhanced Guardian Lattice operations
- Broader ecosystem participation
- SAGE governance framework deployment
Future:
- More specialized domain models
- Enhanced reasoning capabilities
- Broader platform integrations
- Decentralized validator network scaling
Community & Adoption
Beta Phase Success: Thousands of builders used II-Agent during beta to:
- Ship applications
- Conduct research
- Automate complex tasks
Open-Source Community: Growing adoption of:
- II-Agent framework
- II-Researcher
- II-Medical
- II-Thought datasets
Governance Participation: Foundation Coin holders can participate in:
- Sentinel election voting
- Council participation
- Network decisions
Key Strengths
✅ User Sovereignty: Cryptographic key control, no KYC requirement
✅ Transparency: Fully open-source, auditable systems
✅ Governance-First: Guardian Lattice, Oracle Council built-in
✅ Specialized Models: Medical, researcher, reasoning-focused
✅ Production-Ready: II-Agent V1 with strong benchmarks
✅ Decentralization: No single point of failure or control
✅ Alignment Focus: Safety and governance at foundation
Potential Concerns
⚠️ Complexity: Governance architecture steep learning curve
⚠️ Adoption: Smaller community than alternatives
⚠️ Ecosystem Maturity: Newer platform with limited third-party integrations
⚠️ Economic Model: Foundation Coin mechanism still evolving
⚠️ Infrastructure: Scaling decentralized systems is complex
Getting Started
- Explore II-Agent: Available on ii.inc platform
- BYOK Setup: Connect your own API keys (Claude, GPT, Gemini)
- Try Common Ground: Collaborate with AI partners
- Join Community: Participate in open-source development
- Governance: Stake Foundation Coin to participate in decisions
Related Organizations & Concepts
- Emad Mostaque - Founder, AI governance advocate
- Stable Diffusion - Related to founder’s prior work
- Open-Source AI - Broader ecosystem
- Decentralized AI - Governance and infrastructure
- AI Agents - Autonomous systems category
- Sovereign AI - Governance philosophy
Comparisons
- II-Agent vs Agent Zero
- II-Agent vs Claude API
- Open-Source AI Comparison
- Decentralized vs Centralized AI