Platforms where delivery requires per-unit human labor — digital health, care coordination, professional services, marketplace support — can achieve structural, multi-year gross margin expansion when they deploy proprietary AI tools that substitute for routine human interactions. The mechanism is unit-economics-level, not narrative: each automated session reduces clinical/service labor cost per member served. This creates a compounding dynamic: rising automation → rising GM → rising FCF margin → capital available to improve the AI further. The ceiling is higher than pure software peers would suggest — approaching 90%+ in best cases — because the starting floor was artificially low due to human delivery labor.
When evaluating platforms with embedded human delivery labor, identify whether the company has deployed proprietary AI automation for routine interactions. If yes: (1) plot gross margin trajectory quarter by quarter — a consistent 100-200 bps/quarter improvement is the signal; (2) model the ceiling (management guidance or peer benchmark for pure-software comps); (3) discount current GM as understating steady-state — the platform deserves a higher cohort-based EV/S anchor. This pattern is distinct from AI infrastructure cost (which compresses margins) and from SaaS gross margins (which start high). The most actionable companies to screen: digital health, insurance-tech with service layers, HR/benefits platforms, and any B2B SaaS with bundled professional services.