🚀 Get Started: The AI Universe Unfolds 🔗
The AI Big Bang
If 2023 was the AI Big Bang with ChatGPT's explosive release, 2025 feels like First Light.
The fog is lifting, revealing foundational companies and clear patterns for success in the AI cosmos.
Key Statistics
📊 AI Benchmarks: What 'Great' Startups Look Like 🔗
We studied 20 AI startups to understand what separates the exceptional from the ordinary. Two distinct archetypes emerged:
💥 AI Supernovas
Explosively scaling startups with unprecedented growth but often challenging unit economics.
- • ~$40M ARR (Annual Recurring Revenue) in Year 1
- • ~$125M ARR in Year 2
- • Often ~25% gross margins
⭐ AI Shooting Stars
Fast-growing, efficient startups with strong Product-Market Fit (PMF) and healthy margins.
- • ~$3M ARR in Year 1
- • ~$103M ARR by Year 4
- • ~60% gross margins
- • Q2T3 growth pattern
💥 AI Supernovas: Explosive Growth 🔗
These companies achieve unprecedented scale through viral adoption and network effects, often sacrificing margins for growth velocity.
Characteristics:
- • Explosive user acquisition and viral growth
- • High compute costs impacting margins
- • Massive market opportunity capture
- • Often consumer-facing or horizontal tools
⭐ AI Shooting Stars: Efficient Scaling 🔗
These companies follow the Q2T3 benchmark: Quadruple, quadruple, triple, triple, triple growth trajectory with strong unit economics.
Characteristics:
- • Strong Product-Market Fit (PMF) from early stages
- • Efficient go-to-market strategies
- • Sustainable unit economics
- • Often vertical or enterprise-focused
🏗️ AI Infrastructure: The Foundation Layer 🔗
🌟 Forming Galaxies
- ✓ Model layer dominance by big labs
- ✓ Vertical integration trends
- ✓ Reinforcement Learning (RL) tooling emergence
- ✓ Evaluation frameworks maturing
🌑 Dark Matter
The Bitter Lesson
How effective are general-purpose learning methods vs. handcrafted heuristics in embedding context?
Major unresolved questions around the effectiveness of computation and general learning versus domain-specific approaches.
🛠️ Developer Platforms and Tooling 🔗
Model Context Protocol (MCP)
A universal specification for AI agents to access external APIs, tools, and data - like USB-C for AI. Introduced by Anthropic and adopted by OpenAI.
Key Benefits:
- • Persistent agent-to-API connections
- • Standardized data access patterns
- • Cross-platform compatibility
Impact Areas:
- • Memory and context management
- • Tool integration workflows
- • Agent ecosystem development
Memory Systems
Persistent context and learning capabilities
AI Engineering
Integral part of software development
Integration Layer
New infrastructure primitives
🏢 Horizontal and Enterprise AI 🔗
Systems of Action vs. Systems of Record
📊 Legacy Systems of Record
- • Store and organize data
- • Slow implementation cycles
- • Complex data migration
- • High switching costs
⚡ AI Systems of Action
- • Act on existing data
- • Rapid deployment
- • AI-powered wedges
- • Lower implementation barriers
🚀 The AI Wedge Strategy
AI-native platforms can disrupt legacy systems by starting with high-value, low-friction use cases that demonstrate immediate ROI.
🎯 Vertical AI: Industry-Specific Solutions 🔗
Healthcare AI
- • Abridge - Clinical note-taking
- • SmarterDx - Medical coding
- • OpenEvidence - Literature review
Legal AI
- • Contract analysis automation
- • Legal research assistants
- • Document review workflows
Education AI
- • Personalized tutoring
- • Curriculum development
- • Assessment automation
Real Estate AI
- • Property valuation
- • Market analysis
- • Transaction automation
Home Services AI
- • Service scheduling
- • Quality assurance
- • Customer matching
Professional Services
- • Workflow automation
- • Client communication
- • Project management
🔍 Key Success Factors for Vertical AI
Adoption Drivers:
- • 10x productivity gains from day one
- • Seamless workflow integration
- • Industry-specific compliance
- • Domain expertise embedded
Open Questions:
- • Legacy system integration complexity
- • Data privacy and security concerns
- • Regulatory approval timelines
- • Change management resistance
👤 Consumer AI: Personal Intelligence 🔗
Voice & Assistants
AI assistants for task automation, scheduling, and personal productivity.
Creative Tools
Content creation, design, and artistic expression powered by AI.
Health & Wellness
Personalized fitness, nutrition, and mental health support.
🔍 Search Revolution
AI-Native Search
Tools like Perplexity are transforming how we discover and interact with information.
- • Conversational query interfaces
- • Source attribution and verification
- • Context-aware responses
- • Multi-modal search capabilities
Agentic Browsers
Proactive information retrieval and task execution through intelligent browsing.
- • Autonomous web navigation
- • Task completion workflows
- • Personalized content curation
- • Cross-platform integration
🚧 Unsolved Pain Points
Travel & Booking:
- • Complex itinerary planning
- • Real-time rebooking
- • Preference learning
Shopping & Commerce:
- • Personalized recommendations
- • Price optimization
- • Purchase automation
🔮 2025 Predictions: The Year Ahead 🔗
🌐 #1: Browser Dominance in Agentic AI
The browser will emerge as the dominant interface for agentic AI, offering ambient, contextual experiences for autonomous task execution.
Why Browsers Win:
- • Universal access to web services and APIs
- • Contextual awareness of user workflows
- • Cross-platform compatibility
- • Existing user behavior patterns
🎬 #2: The Year of Generative Video
2026 will be the breakthrough year for generative video, with mainstream adoption in entertainment and marketing.
📊 #3: Evals Catalyze Development
Private evaluation frameworks and data lineage will become the catalyst for trusted AI product development.
📱 #4: AI-Native Social Media Giant
A new social media platform will emerge, built from the ground up with AI agents, voice interfaces, and generative capabilities.
🤝 #5: M&A Acceleration
Incumbents will aggressively acquire AI-native startups to catch up, especially in vertical software markets.
💡 The Founder's Edge in the AI Cosmos 🔗
Key takeaways for AI founders navigating the evolving landscape and building defensible, scalable businesses.
🧠 Build Memory Moats
Memory and context are becoming the new defensibility. Focus on systems that learn and remember user preferences, workflows, and domain knowledge.
- • Persistent user context across sessions
- • Domain-specific knowledge accumulation
- • Personalized model fine-tuning
- • Workflow pattern recognition
⚡ Focus on Systems of Action
Don't just store data—act on it. Build AI-native platforms that can execute tasks and make decisions, not just provide insights.
- • Autonomous task execution
- • Real-time decision making
- • Workflow automation
- • API-first architecture
🔧 Start with AI Wedges
Identify high-friction, high-value problems where AI can deliver immediate 10x improvements. Use these as entry points to larger markets.
- • Language-heavy workflows
- • Multi-modal data processing
- • Repetitive expert tasks
- • Complex decision trees
📊 Implement Private Evals
Build proprietary evaluation frameworks on your specific use cases and data. This creates trust, enables rapid iteration, and provides competitive advantage.
- • Use-case-specific metrics
- • Proprietary test datasets
- • Continuous feedback loops
- • Data lineage tracking
🎯 Strategic Framework for AI Founders
❓ Frequently Asked Questions 🔗
📝 Original Content Contributors
Partners
- • Kent Bennett - AI & Developer Tools
- • Talia Goldberg - Consumer Tech & AI
- • Janelle Teng - AI Infrastructure & Vertical AI
- • Sameer Dholakia - SaaS & AI
- • Mike Droesch - AI & Machine Learning
Investors
- • Lauri Moore - AI & Healthcare
- • Caty Rea - Consumer AI
- • Lindsey Li - Fintech & AI
- • Maha Malik - AI & Data
- • Bhavik Nagda - AI Infrastructure
Contributors
- • Lance Co Ting Keh - Developer Platforms
- • Darsh Patel - AI & Cloud
- • Christine Deakers - Content Lead