Source: Trading Margin for Moat / Services-Led Growth (a16z)

Joe Schmidt's 2025 argument that complex enterprise AI apps should accept services-heavy implementation (FDEs) early, as Salesforce, ServiceNow and Workday did in the cloud shift.

Contents

Summary

Argues that thin, wrapper-style integrations fail to replace core enterprise workflows, and that AI application companies should embrace implementation work, often done by FDEs, to become deeply embedded. Uses cloud-era SaaS margins to show that services-heavy companies can grow into software margins.

Key claims

  • FDEs (also called implementation or solutions specialists) connect the AI product to internal systems and data and supply the business context agents need, much like onboarding a new employee.
  • Model providers such as OpenAI and Anthropic hire FDEs and solutions engineers (22 of 311 open OpenAI roles at the time of writing); Decagon uses “Agent Product Managers” to configure its support agents.
  • Gross margin at IPO was 63.2% for ServiceNow and 54.1% for Workday, against roughly 80% for typical software; by 2024 they reached 79% and 75%.
  • Early-stage AI companies should optimise total gross profit rather than margin percentage.
  • Deep integration makes products hard to replace and creates pricing power; the goal is to become the “system of work” where data is created.
  • AI can automate parts of implementation itself, making the services model cheaper than in the cloud era.
  • Advice: focus on a customer segment, align services with sales, build reusable APIs and docs, automate implementation tasks.

Feeds

Forward deployed engineering

Suggest changes →