The challenge for enterprises deploying artificial intelligence (AI) is increasingly not just selecting or building a model, but integrating it with legacy systems, adapting it to business workflows and running it reliably in production. A study by CyberMedia Research (CMR) found that 74% of enterprise respondents identified AI and generative AI (GenAI) customisation as an area requiring Forward Deployed Engineering (FDE) capabilities, while 63% pointed to legacy and enterprise-system integration.
The findings, from a study commissioned by technology education company Scaler, highlight the engineering work required to move AI projects beyond proofs of concept and into operational use.
FDEs typically work at the intersection of software engineering, AI implementation and customer requirements. Rather than developing a model in isolation, they help adapt technology to an organisation's data, applications and processes, and address problems that emerge during deployment.
The survey covered 200 enterprise hiring decision-makers and more than 1,000 engineers in September 2026.
From AI pilots to production systems
Moving an AI pilot into production involves more than demonstrating that a model can perform a task. Enterprises must connect it to internal data and applications, manage access permissions, test its outputs and ensure that it performs reliably within existing workflows.
In the CMR study, 55% of enterprise respondents identified moving pilots into production as an area where FDEs could create greater value. This points to the gap between testing AI in controlled environments and making it work within business operations.
For example, an AI assistant designed to support customer service may need access to customer records, integration with a customer relationship management platform and safeguards against exposing sensitive information. Engineers must ensure that the components work together and that the resulting system meets operational requirements.
This work can involve application programming interfaces (APIs), data pipelines, model integration, monitoring and troubleshooting. It also requires an understanding of the business problem the system is intended to solve.
Technical skills remain a constraint
The report identifies a gap between the capabilities enterprises value and the skills they believe are available in the market. While 77% of enterprise respondents considered production deployment an important FDE capability, only 19% identified those skills as readily available.
The findings suggest that employers are looking for more than AI model development. Engineers need to understand deployment environments, integrate AI with enterprise applications and resolve issues that arise when systems encounter real-world data and workflows.
Customer engagement is another part of the role. Requirements often change during implementation, making it necessary for engineers to translate business needs into technical decisions and work with users to validate the results.
The report's findings also point to demand for AI customisation and enterprise integration, areas where solutions must account for organisation-specific data, infrastructure and processes.
What it means for enterprise AI
The growing interest in FDE capabilities reflects a broader issue in enterprise AI adoption: a technically capable model does not automatically translate into a useful business application.
Organisations must also address data quality, security, access controls, system compatibility and ongoing monitoring. These requirements become particularly important when AI is connected to business-critical applications or given the ability to trigger actions.
FDEs could help bridge the gap between AI development teams and the business units using their systems. However, the role overlaps with existing functions, including solutions engineering, implementation consulting and AI engineering. Its long-term significance will depend on whether organisations establish it as a distinct role or distribute its responsibilities across existing teams.
The Scaler-commissioned study provides an indication of what surveyed employers want, but it does not establish the size of the overall FDE market or confirm future hiring volumes. Independent enterprise perspectives will be needed to determine how widely the role is being adopted.
For technology leaders, the central question is not simply whether to hire FDEs. It is how to build the engineering capabilities needed to integrate AI into existing systems, maintain reliability and deliver measurable business outcomes.