The term is not new, with Palantir claiming to have coined it in 2007, but investment is now growing explosively. The context is stark: Gartner estimates the industry spent around $1 trillion on AI last year and expects that figure to grow 34.7 percent in 2026, with vendors hoping to claw money back from customers. Salesforce chief Marc Benioff told an earnings call in December last year that customers working with its FDEs deployed the company's AI technology 33 percent faster than those who did not. Pay reflects demand: Levels.fyi data puts the US median annual total compensation for FDEs at $216,250, ranging from $180,000 at the 25th percentile to $290,000 at the 75th.

The risks for buyers are real, according to Gartner. Mukul Saha, a software engineering senior director at the firm, warned that if FDEs become indispensable because only they understand the workflows they built, keeping them on as full-time consultants is neither economical nor strategic. He advised customers to define use cases before an engagement and to plan an exit strategy covering how they will take over once FDEs leave. Saha also flagged IP concerns, noting a fine line between vendors learning from implementations to improve their products and passing a customer's competitive advantage to its close rivals, a worry he said he hears from end user clients.

Licensing is another blind spot. Craig Guarente, founder and CEO of advisory firm Palisade Compliance, said FDEs generally do not check licensing, tokenisation or what customers are paying for, leaving buyers exposed if contracts are open-ended or usage is capped and limits get hit. Gartner's advisory research does offer a positive path: with clear guidelines on governance, business value delivery, IP ownership, knowledge transfer and an exit strategy, FDE engagements can get projects to production faster and improve the chances of a return on investment.