Directions Brain

Services: The New Software

3 min read

Services: The New Software

Source: raw/articles/vc-articles/2026-03-05-sequoia-services-the-new-software.md

Note: captured from the public article URL using direct curl + pandoc fallback because the axcli chrome gateway was unavailable in this session. The raw file preserves the article body and provenance block.

Summary

julien-bek of sequoia-capital argues that the next generational AI company may look like a services firm rather than a classic software vendor. The core distinction is copilot versus autopilot: copilots sell tools to professionals, while autopilots sell completed work directly to the customer who needs the outcome. The article’s operating playbook is to start with outsourced, intelligence-heavy work where the customer already buys an outcome, then expand toward insourced and more judgment-heavy work as proprietary domain data improves the system.

Key claims

  • Selling work is structurally stronger than selling tools. Tool vendors race against model improvement, while service/outcome vendors can turn model improvement into lower cost, faster delivery, and better margins.
  • Intelligence-heavy work automates first. Sequoia separates rule-bound execution from judgment. Software engineering crossed the threshold first; the article argues similar thresholds will appear in other professions.
  • Autopilots can capture work budgets immediately. When the buyer purchases an outcome rather than seats or workflow software, the addressable budget shifts from software spend to labor and services spend.
  • Outsourcing is the preferred wedge. Existing outsourcing proves the work can be externalized, creates a substitutable budget line, and makes the AI-native service provider a vendor swap rather than an internal reorg.
  • The long-term TAM is insourced work. The initial outsourced task should be narrow and intelligence-heavy, but the end state is broader labor substitution as the product accumulates domain data and pushes into judgment-heavy work.
  • Copilots and autopilots will converge. Copilots have product/customer knowledge, but moving to services can cut against their existing customers; that creates room for pure-play autopilots.
  • Best initial markets have large services spend plus high intelligence ratio. The opportunity map names insurance brokerage, accounting/audit, healthcare revenue cycle, claims adjusting, tax advisory, transactional legal work, managed IT, supply chain/procurement, recruitment/staffing, and selected consulting components.

Hypotheses raised

Notable quotes

  • “A copilot sells the tool. An autopilot sells the work.”
  • “For every dollar spent on software, six are spent on services.”

Mentioned in

Graph View

Backlinks