Services: The New Software

· Source ↗

TLDR

  • AI autopilots that sell outcomes capture the services budget, which is 6x larger than the software budget, leaving tool-sellers racing against the model.

Key Takeaways

  • Copilots sell tools to professionals; autopilots sell outcomes to buyers – Crosby sells NDAs to companies, not outside counsel.
  • Outsourced tasks are the wedge: budget exists, scope is clear, substitution is a vendor swap not a reorg.
  • Software engineering leads AI adoption at 50%+ of all AI tool usage across professions; every other category is still single digits.
  • Largest autopilot markets by dollar: management consulting $300-400B, staffing and procurement each $200B+, IT managed services $100B+.
  • Accounting faces a structural supply crisis: US lost 340,000 accountants over five years; 75% of CPAs are nearing retirement.

Why It Matters

  • In 2026, fast-growing copilot companies face the innovator’s dilemma: selling the work means cutting their own customers out, opening the door for pure-play autopilots.
  • The playbook is to wedge on outsourced intelligence-heavy tasks, then expand into insourced judgement-heavy work as proprietary domain data compounds.
  • Today’s judgement becomes tomorrow’s intelligence as AI accumulates domain-specific data, meaning copilots and autopilots will eventually converge.

Julien Bek, Sequoia Capital — March 5, 2026 · 2026-03-06 · Read the original