The Hidden Risk of Vendor-Deployed AI Engineers

Software vendors are aggressively deploying "forward-deployed engineers" (FDEs) to client sites to accelerate AI implementation, as The Register first reported. These specialists, described by the outlet as "consultants-without-portfolio," are tasked with building valuable systems at speed using the vendor's proprietary technology.

Major players are investing heavily in this model. AWS recently announced a $1 billion investment in a new FDE unit, while Anthropic has committed $100 million to training these engineers. Other firms heavily invested in the strategy include Salesforce, SAP, and Workday, the latter in partnership with Accenture. Salesforce CEO Marc Benioff claimed during a December earnings call that customers using FDEs deployed AI technology 33% faster.

However, this rapid deployment comes with significant operational risks. Mukul Saha, software engineering senior director at Gartner, told The Register that relying on FDEs for every iteration can create "enormous lock-in." If FDEs become indispensable because they are the only ones who understand the built workflows, the arrangement becomes neither strategic nor economical. Gartner is advising clients to define their use cases and exit strategies before FDEs arrive on site.

Beyond lock-in, there are concerns regarding intellectual property. Saha noted that FDEs may take knowledge gained during implementation back to the vendor to improve the product, risking the leak of a customer's competitive advantage. Additionally, Craig Guarente, CEO of Palisade Compliance, warned that FDEs typically do not monitor licensing or tokenization. Consequently, buyers with capped usage or open-ended contracts may find themselves "really exposed" if they do not strictly manage the commercials and software usage themselves.