The AI Opportunity that goes beyond Models

• AI Era Identification: The current market is defined as the "AI era," building upon prior infrastructure layers (PC, Internet, Cloud, Mobile) and characterized by 8 billion smartphone users enabling rapid adoption.
• Revenue Attribution: Net new revenue in the software sector is increasingly driven by AI at both the application and infrastructure layers.
• Behavioral Driver: The primary adoption catalyst is a dual human desire to be "richer and lazier," with Gen AI enabling reduced labor for increased economic value.
• Growth Metrics: Software companies are achieving unprecedented growth, scaling from zero to $100 million in revenue within one to two years, a pace not observed in previous software cycles.
• Adoption Data: Approximately 15% of adults globally use ChatGPT weekly for daily routine tasks, with usage minutes growing at an astronomical rate.
• Technological Foundation: The 2017 "Attention is All You Need" paper introducing the Transformer model marked the shift from early, ineffective AI (e.g., ELIZA-style chatbots) to current "golden age" capabilities.
• Enterprise Validation: Forward-thinking enterprises (e.g., Ramp) are adopting AI earlier than traditional incumbents, driving immediate efficiency gains in expense management and operations.

Investment Themes in AI Applications

• Theme 1: Traditional Software Going AI Native

  • Greenfield Opportunities: Highest success rates for new entrants occur when targeting "Greenfield" markets (new companies or inflection points like multi-entity ERP needs) rather than displacing incumbents in "Brownfield" markets.
  • Incumbent Resilience: Existing incumbents (e.g., Adobe, Workday, NetSuite) are leveraging AI to strengthen "hostages" (sticky systems of record) rather than being displaced.
  • Pricing Models: Shifts are occurring from per-seat pricing (e.g., Zendesk) to outcome-based pricing where 99% of queries are handled by AI, requiring new monetization strategies.
  • System of Record: The most defensible companies become the central "system of record" for business operations, making displacement prohibitively difficult.

• Theme 2: Software Earning Labor (Labor-as-Software)

  • Market Size: This category represents a market significantly larger than traditional software because it directly replaces human labor rather than augmenting tools.
  • Value Proposition: Companies selling "labor replacement" can charge $20,000 annually for software performing 5 of 8 job responsibilities, capturing value previously reserved for salaries.
  • Use Case Example (Eve): A legal AI platform for plaintiff attorneys operates on a contingency model, using AI voice agents to gather evidence and prioritize cases, creating a data moat that improves intake accuracy.
  • Use Case Example (Salient): Auto loan collection software collects 50% more revenue than human teams by handling 24/7 calls in 21 languages and adhering to complex, real-time statutes, solving a labor shortage issue.
  • Moat Strategy: Defensibility is derived from owning the end-to-end workflow and proprietary outcome data, not just the underlying AI primitives.

• Theme 3: The "Walled Garden" (Proprietary Data)

  • Data Moats: Companies that aggregate free but unorganized public data (e.g., flight transponders, legal records, domain history) and digitize them create exclusive "vegetable farms" that infrastructure providers cannot replicate.
  • Finished Products: The value lies in transforming raw data into a "finished meal" (e.g., Open Evidence for medical journals, V-Lex for Spanish law), allowing for premium pricing over raw data subscriptions.
  • Incumbent vs. Startup: While incumbents like Bloomberg and LexisNexis possess data, startups are creating new walled gardens by digitizing previously inaccessible or unstructured historical data (e.g., old blender manuals, YouTube subscriber history).
  • AI Amplification: AI increases the valuation of these data assets by enabling the delivery of high-value, specific insights that were previously too expensive to produce manually.

Strategic Dynamics and Market Evolution

• Differentiation vs. Defensibility: Voice or summarization features offer differentiation but are not defensible; defensibility requires becoming a system of record with proprietary, non-public data workflows.
• Aggregator Trend: In consumer AI, aggregators (e.g., Kayak for flights) will likely win over single-model providers because users prefer access to a "single pane of glass" for multiple specialized models.
• Incumbent Response: Unlike the Cloud era where incumbents were initially skeptical, AI is widely recognized as beneficial, leading incumbents (e.g., Intuit/QuickBooks) to aggressively monetize their existing "hostage" customer base.
• Roll-Up Strategy: Vertical roll-ups of professional services (e.g., accountants, dermatology) face high customer acquisition costs; AI roll-ups are more viable when they acquire companies with existing "blue-chip" clients and transform operations with AI to reduce churn and increase efficiency.
• Retention Factors: Enterprise retention is strong as long as startups build rich ecosystems around AI primitives; customers view startups as holistic solutions for AI integration rather than point tools.
• Sales Motion: Enterprise sales for AI startups increasingly rely on "forward-deployed engineering" to help large corporates understand application use cases, rather than traditional outbound sales.

Andreessen Horowitz (a16z) Operational Approach

• Deal Sourcing: The firm utilizes an "interruption" strategy where experts drop all tasks to pursue high-conviction deals, prioritizing "positive selection" (meeting the best companies) over "adverse selection" (saving time on low-quality deals).
• Investment Process: A "two-key" system ensures high conviction, where the lead partner makes the decision subject to rigorous process adherence rather than committee voting; the firm defers to the partner with deep domain expertise.
• Team Evolution: The firm is adding senior partners with "quasi-generational" company-building experience to increase leverage in winning "superpower" deals, rather than simply adding capacity to find more deals.
• Content Strategy: Publishing industry-specific content (e.g., "Death of a Salesforce," AI productivity benchmarks) serves as a primary tool for deal generation and establishing category expertise.
• Consumer Application: The firm applies the same three themes to consumer AI, citing examples like Creo (AI-native design), Eleven Labs (voice creation), and Slingshot (AI therapy using proprietary data from therapist scribes).
• Future Outlook: The firm anticipates that while AI will displace specific tasks (e.g., truck driving, data collection), it will primarily augment labor by inverting the value/cost equation, enabling humans to focus on tasks requiring higher judgment or emotional intelligence.