Why AI Giants are Outsourcing Forward Deployed Engineering

Enterprise AI adoption is hitting a wall of integration complexity, and the industry's biggest players are responding by spinning off their deployment teams into quasi-external organizations.

The shift toward outsourced deployment models

Google Cloud CEO Thomas Kurian recently announced a new, AI-focused organization within the Go-To-Market team, specifically targeting a spike in Forward Deployed Engineer (FDE) recruitment. The urgency is visible in the hiring process, which has reportedly been compressed from weeks of interviews down to just two interviews held over two days.

OpenAI is following a similar trajectory by funding The OpenAI Deployment Company, a standalone entity backed by $4 billion in private equity from TPG and Advent at a $14 billion valuation. This move allows OpenAI to focus on core model development while the Deployment Company handles the heavy lifting of redesigning organizational infrastructure and critical workflows around AI. This strategy includes the acquisition of Tomoro, a UK-based AI company that employs 150 FDEs across the UK, Asia, and Australia.

Anthropic is also moving toward this model, issuing plans to create a distinct FDE consulting company backed by investors including Blackstone, Hellman & Friedman, and Goldman Sachs. The goal is to integrate Claude into the core operations of mid-sized companies, effectively driving higher token consumption through deep enterprise integration.

From platform engineering to high-stakes consulting

The nature of the FDE role is undergoing a fundamental transformation. A year ago, FDEs at companies like OpenAI and Ramp functioned as a hybrid of platform engineering and solutions engineering, contributing code back to the core product while building customer-facing solutions.

Today, the role is increasingly indistinguishable from a systems integrator or a high-end consultant. Job descriptions, such as those recently posted by Google Cloud, describe FDEs as "innovator-builders" who move beyond high-level architecture to code, debug, and ship bespoke agentic solutions directly within customer environments.

However, the reality of the work is shifting toward the "plumbing" of AI. For a typical FDE, the workload is estimated at roughly 25% coding, 50% integration and data readiness, and 25% customer management and meetings. The focus has moved from building the platform to solving the state-management challenges and integration complexities that prevent AI from reaching enterprise-grade maturity.

This evolution suggests that as AI models become commoditized, the real value—and the real engineering struggle—will lie in the specialized layer of deployment and workflow redesign.

Source: The Pulse: Forward deployed engineering heats up again
Domain: blog.pragmaticengineer.com