- Kimi, Tencent Cloud and 01.AI have each launched distinct FDE initiatives, signaling that the role has moved from Silicon Valley trend to mainstream Chinese AI industry practice.
- Practitioners interviewed by QbitAI report spending over 70 percent of their time on communication and organizational alignment rather than coding, describing enterprise AI deployment as a change-management challenge.
- Independent guerrilla FDEs are emerging as a new category, with solo practitioners completing dozens of projects across banking, manufacturing and services, often pivoting from building software to training employees.
- The economics of FDE work are being reshaped by AI itself: development costs have collapsed, but the cost of understanding client organizations and integrating with legacy systems remains stubbornly high.
- Multiple practitioners describe FDE as a transitional role whose ultimate purpose is to make itself unnecessary by improving products, though rapid AI evolution may keep the gap perpetually open.
From Silicon Valley to Chinese Enterprises
QbitAI's October 4 report describes how the Forward Deployed Engineer role has arrived in China's AI industry. The report cites Anthropic's USD 100 million training commitment, OpenAI's USD 4 billion deployment venture and AWS's USD 1 billion FDE department as international context. Major domestic players are adapting the model for a market where many enterprises cannot yet articulate where AI should be applied.
Kimi (Moonshot AI) has announced partnerships with multiple IT service providers to jointly build FDE teams. Tencent Cloud has launched what it describes as the industry's first FDE engineer certification program while actively recruiting FDE partners. Li Kaifu's 01.AI has fully pivoted toward enterprise AI. QbitAI describes an early project involving five on-site FDEs and five backend engineers; Li Kaifu says building a client's Ontology can take one to three months.
Corporate FDEs Versus Guerrilla Practitioners
The Chinese FDE landscape has bifurcated into two distinct populations. Corporate FDEs work within product organizations, embedding at client sites for weeks or months to map business processes, build prototypes, and feed insights back into standardized products. Their value proposition depends on accumulating reusable modules in Skill Hubs and Agent Hub platforms that compress delivery timelines from months to days for subsequent engagements in the same vertical.
Outside the corporate structure, independent practitioners have emerged as what the community calls guerrilla FDEs. One Stanford doctoral student completed dozens of enterprise AI projects across banking, law firms, coal-chemical plants and maternity centers in just five months, working entirely alone. His business model has shifted from building custom software to functioning as an external chief AI officer, teaching employees to use existing tools and connecting them to internal systems through lightweight integrations.
The Real Bottleneck Is People, Not Technology
Several of the practitioners interviewed by QbitAI said they spent more than 70 percent of their time on communication and roughly 30 percent on development. The report describes difficulties obtaining API access from reluctant IT departments, aligning expectations about what AI can realistically achieve, and navigating organizational concerns when automation threatens existing roles.
One practitioner recounted a client leader's response to suggesting a general-purpose agent for certain modules: if you use WorkBuddy for everything, what will our people do? This encapsulates the core tension — the cost of changing how an organization works often exceeds the cost of the technology itself, and no amount of coding efficiency eliminates the need for trust-building and stakeholder management.
AI Rewrites the Economics but Not the Organizational Cost
Traditional software deployment followed a model where development was expensive and per-client adaptation was relatively cheap, allowing upfront costs to amortize across customers. AI has inverted this equation: generating code is now nearly free, but each new enterprise engagement still requires fresh effort to understand workflows, connect legacy systems, secure permissions and persuade different departments to adopt new practices.
This structural reality raises questions about whether FDE work can truly scale. While individual Skills and industry knowledge are transferable, the human-intensive portions of the job do not benefit from the same exponential cost reductions as code generation. Several practitioners acknowledge that FDE may ultimately be a transitional function necessary only because products are immature and organizations are unprepared, but argue that the pace of AI evolution means the gap between product readiness and organizational readiness may never fully close.
Sources & context
Go to the original material. Company claims remain attributed to their sources.
01
