Interview, Fireside Chat
The FDE Playbook for AI Startups with Bob McGrew
- The Forward Deployed Engineer (FDE) model is expected to become the dominant organizational structure for AI agent startups over the next five years, driven by the necessity of continuous product discovery in a heterogeneous market with no incumbent products.
- FDE-led discovery is predicted to outperform traditional sales-led approaches by solving problems from the inside, enabling startups to "land and expand" within enterprises by identifying key problems beyond initial top-five priorities.
- Contract sizes in the FDE model are projected to grow larger and more flexible over time, contrasting with standard SaaS models, with startups expected to assume early risks and offer performance-based pricing to overcome enterprise skepticism.
- Profitability is anticipated to take one to multiple years to achieve as early deployments result in initial losses that decrease only as products become better suited to customers and the cost per value of the outcome declines.
- The market is projected to evolve from specific single-customer products to a customizable platform serving multiple segments, with a historical trajectory similar to Palantir's shift from file-based to database-driven collaboration.
- A significant number of startups are currently learning the FDE strategy, with Y Combinator startups hiring for these roles expected to have grown from zero to over 100 within the three years prior to the conversation.
- The FDE role functions as "product discovery," where engineers build initial "gravel road" solutions for one customer that product teams later generalize into "paved superhighways" for the next five to ten customers.
- Operational teams are expected to consist of domain experts ("Echo") and rapid prototypers ("Delta") who write disposable code, while the model aims to balance immediate customer-specific solutions with long-term abstraction and maintainability.
- Startups lacking Palantir alumni in core FDE roles are expected to struggle with outcome selection and pricing mechanics, while the presence of such personnel provides significant benefits in navigating execution capabilities.
- On-premise deployments require specific executive buy-in to bypass IT teams not aligned with startup agility, and success metrics are expected to shift from minimizing custom work to increasing the value of the delivered outcome.
- Rapid AI capability progress is projected between April 2024 and April 2025, creating a massive gap between technical capabilities and customer adoption that startups must bridge over the next five years.
- The industry may undergo a significant definition shift where the current concept of "AI agents" is viewed retrospectively as nonexistent within five years, as the market moves toward bridging capability with genuine human utility.
- Product leverage is expected to increase over time, allowing FDEs to deliver more value without additional engineering resources, though this requires building unrequested features if they provide necessary scale.
- Demo-driven development is expected to become a superior strategy by forcing a customer-centric perspective, while companies are projected to maintain "learning" cultures to avoid the inertia of success.
- The application of FDE strategy logic is expected to extend to the U.S. Army under Bob McGrew to address the disconnect between leadership priorities and legacy implementation methods spanning 20 years.