Video summary

There Is No Demand For Average — Naval Ravikant on AI & Wealth

Main summary

Key takeaways

Business

Core thesis: “Average” is being erased by AI, and pure software advantages are shrinking

  • AI is rapidly automating not only coding, but the creation of software assets and even agents that can build and/or operate systems.
  • This compresses timelines: advantages (models, tooling, features) become obsolete quickly as capabilities are copied or commoditized.

Implication for strategy

Compete on:

  • Distribution and partnerships
  • Hardware/positioning that buys time
  • High-touch customization rather than on “unique software code” alone.

Why “pure software” is becoming less investable (moat dilution)

  • VCs increasingly worry about software lock-in; if code can be generated quickly, differentiation collapses.
  • “Coding agents” and AI-assisted development reduce barriers to replicating software products.
  • Cloud/code commoditizes everything: if something can be specified, AI can generate it in one-shot or via quick iterations.

New product/market opportunity: individual creators and small teams

AI lowers the cost of building products—so many niches can be served by builders without a “middle layer.”

Small teams can scale significantly because:

  • AI can handle coding, iteration, bug fixing, and customer responses.
  • Teams can run more experiments and ship faster, creating feedback/reward loops similar to game design.

Custom + private “apps” vs best-of-breed general apps

Likely direction:

  • More device- or account-specific customization
  • Potentially more private deployments

Strategy split

  • General apps (broad use cases): remain “best of breed” and can win through quality and breadth.
  • Niche/custom apps: win when value comes from tailoring, privacy, or highly specific workflows.

Org/leadership impact: AI changes talent leverage and hiring profiles

  • Productivity rises with AI.
  • Economics implies you may hire more top people rather than fewer.

Suggested hiring direction

Move toward “juniors and super seniors”, with roles emphasizing:

  • creativity
  • effective AI use
  • steering outcomes rather than only writing code

Debate shift

The balance of advantages shifts toward agency (execution with AI) vs “pure intelligence”-heavy advantage (exact ratio contested).

Human role: humans become verifiers + taste/judgment operators

AI is positioned as:

  • implementer/assistant (agents follow instructions)

So the human becomes a verifier, responsible for:

  • confirming correctness
  • managing risk
  • stepping in when things go wrong

What humans still uniquely provide

  • Taste and judgment (e.g., choosing the right telemetry system/storage approach)
  • Motivation and desire (still critical for product direction and user acceptance)

Frameworks / playbooks referenced

  • Moat via timing (time-buying)
    • Hardware can “buy time” for software moats, but AI/cloud commoditize software quickly.
  • Feedback loop principle (game design analogy)
    • Design workflows where users get continuous reward/feedback to improve retention and iteration speed.
  • Human-in-the-loop verification model
    • Shift operational burden from “do everything” to verify and steward outputs in production.
  • Talent leverage model (skill mix)
    • As AI handles coding, humans steer goals and enforce quality thresholds (agency becomes more dominant).

Concrete examples / case patterns mentioned

  • VC diligence question
    • “What’s the software lock-in?”—especially after hardware funding.
  • Architecture taste example
    • AI suggesting where to place high-cardinality telemetry (humans rejecting/choosing better tools), e.g. ClickHouse/Athena vs Postgres.
  • Historical scaling examples
    • Small teams producing outsized impact (early Instagram, early WhatsApp; also cited: Notch; Satoshi Nakamoto).

Metrics / KPIs and targets

  • No explicit financial or growth targets were given (e.g., revenue, CAC, LTV, churn).
  • Indirect operational “metrics” referenced:
    • Productivity increase (“productivity has gone through the roof”)
    • Timeline compression (“within a year or even less,” “2 weeks, 3–4 weeks”)
    • Token cost as a proxy for compute efficiency (presented as speculative intuition; no hard KPI target provided)

Actionable recommendations (derived from the business arguments)

  • Don’t pitch “cool unique software” as the core moat. Assume AI can replicate it quickly.
  • Build defensibility via:
    • distribution and partnerships
    • hardware/positioning that buys time
    • proprietary data and workflow entrenchment
    • customization/agents that are hard to generalize into a one-shot product
  • Adopt human verification operations:
    • treat AI outputs as drafts
    • implement review/checking workflows
    • define correctness/taste gates for production
    • shift legal/ops toward verification and accountability
  • Build org capability around AI workflows:
    • hire for creativity + agency with AI
    • cultivate “taste loops” where judgment improves output quality over time
  • Run more experiments:
    • AI reduces iteration cost; capitalize with faster shipping and learning cycles

Presenter / sources

  • Naval Ravikant (main subject)
  • Other speakers/participants referenced in subtitles as “Max” and other unnamed interview/conversation participants.

Original video