"I thought that becoming an AI product manager would keep my job safe from AI for a while. But looking at it now, that was too naive. AI won't let you off easily—if you want to survive longer, you should keep transforming. My current goal is to become an FDE."
As AI deployment turns into real productivity, its substitution effect has begun to show. Product manager roles that lean toward execution and are highly functional sit at the center of the "eye of the storm," triggering anxiety within this group and spurring a wave of career transformation attempts.
However, the path of transformation is not an easy one.
said a product manager who has moved across consulting, consumer goods, SaaS, and other industries.
With a track record of independently owning a product line, a computer science degree, SaaS experience, and constantly updated AI knowledge born of hard work, he landed an AI product manager job last year, which was seen as a career step up, and his salary did indeed increase. But this year, the explosion of AI capabilities has dramatically shortened development cycles, while companies increasingly focus on whether products can actually make money. His role is "squeezed from both ends," and the anxiety of being replaced has begun to attack him again.
"As an 'older' PM, if your results are unsatisfactory, you get told you're not AI Native enough. From that angle, becoming an FDE might indeed be a better fit." Recently, he turned his attention to an even trendier role: FDE (Forward Deployed Engineer).
This role is usually seen as an on-site expert who helps industry customers deploy AI. In fact, until 2025, FDE remained fairly niche. But over the past year or so, from Silicon Valley to China, from model unicorns to cloud providers, demand for FDEs has surged, making it one of the most-watched new roles in AI commercialization.
The shortage of matching talent, path dependence on old habits, and mismatches in fit have, on one hand, created bottlenecks in the actual rollout of FDE roles, and on the other, opened a new transformation path for many "workers" caught between the traditional and AI eras.
The Interface
FDE is not a native concept of the AI era.
As early as the beginning of the 21st century, big data giant Palantir pioneered the FDE model. At the time, the core demand came mainly from the U.S. military, government, intelligence agencies, and defense contractors that this "mysterious company" served. Palantir could not adapt to these highly confidential, highly mobile, low-tolerance-for-error customer scenarios with hard-to-articulate needs using the traditional "requirements research—software development" model, so it sent engineers directly to the field to write code alongside customers.
Although Palantir reaped the benefits of FDE and became increasingly popular among high-demand, high-premium customers such as the military, government, and large enterprises, from a business model standpoint, FDE runs almost counter to the SaaS "software as a service" model: high deployment costs, difficulty standardizing, weak economies of scale, and slow declines in marginal costs. For a long time, few other software companies followed suit.
It was not until this wave of AI surged forward and swept through the business processes of industries and enterprises that FDE finally exploded.
In April this year, job postings for FDE roles on the recruiting platform Indeed reached 5,330, up more than sevenfold year over year, with salaries generally between $150,000 and $300,000, and some companies offering senior FDEs more than $500,000.
Screenshot from Indeed
AI model vendors and cloud giants have become the main drivers of FDE. A landmark event was OpenAI's announcement in mid-May of the establishment of DeployCo, an organization focused on making AI on-site deployment organized, scaled, and productized, with initial funding of over $4 billion. It quickly acquired AI consulting firm Tomoro, bringing in about 150 FDEs at once. Other leading AI companies, including Anthropic, Google, and Microsoft, are also increasing their investment.
On July 6, well-known Silicon Valley AI investor Tomasz Tunguz wrote that over the past year, mainstream AI companies had invested about $10 billion in forward deployed engineering, an investment that is making more people understand: without people who can handle installation, AI doesn't work.
In Silicon Valley job descriptions for related positions, FDEs are typically required to have highly proactive communication skills, the execution ability to deliver measurable business impact from day one, the capacity to work side by side with customer team members, rapidly deliver prototypes, and provide viable AI solutions—emphasizing producing visible results on-site at the customer. This fits exactly with the expectations, during the AI deployment phase, of both capital and market supply and demand for AI to deliver real economic value.
In other words, FDEs are expected to become the "human interface" connecting cloud-based AI capabilities to on-the-ground business sites.
The Gray Zone
Chinese companies are also following the FDE trend, with core demand coming from major internet companies and leading AI model makers.
According to reports, on all major mainstream recruiting platforms, postings for FDE-related positions grew tenfold or twentyfold in the first half of this year. Salaries are also considerable: as of early July, there were nearly 300 FDE positions on the Liepin platform with total annual salaries between 500,000 and 1 million RMB, and no fewer than 40 with annual salaries exceeding 1 million. A headhunter had just posted an FDE leader position for a well-known AI company in Shenzhen in early July, mainly responsible for deploying AI in manufacturing sectors such as robotics and new energy, with a maximum annual salary of over 3 million. Entry-level FDE positions at ByteDance, Tencent, Alibaba, and others generally start at over 20,000 RMB per month, with senior roles mostly at 40,000/month × 15 months or above. Reportedly, actual hiring varies considerably depending on candidates' track records, deployment locations, and customer needs, and there are even cases of non-leader technical roles with annual salaries exceeding 1 million.
Screenshot from Liepin
Beyond mainstream AI companies, consulting firms, chip companies, traditional IT companies, and others are releasing more and more related positions. And while the requirement of understanding both business and AI seems high, it is precisely the absence of "native talent" and matching disciplines that has created a "gray zone" between job supply and demand. Programmers, product managers, and digital transformation consultants skilled in implementation and deployment engineering form the main applicant pool, and on the management track, including some consulting project managers with technical backgrounds and solutions sales talent, have also been able to get past the threshold.
The aforementioned product manager planning to transform into an FDE said that during his preparation and job search, he found that cross-industry experience plus AI development experience can be an advantage for FDEs. "Many job postings mention deep expertise in a certain industry, but how long counts as deep? In fact, for an FDE earning over 300,000 RMB a year, two years of full-cycle experience in development, delivery, and implementation in a given field is enough."
He believes that beyond basic conditions like vertical industry track records, AI development skills, and the ability to be on-site long-term or travel frequently, what companies value most is dealing with customers—the ability to independently uncover needs, summarize pain points and blockers, and distill standardized methodologies. To some extent, this suits product managers with technical backgrounds and experience. On the community of renrenshidouchanpinjingli.com (PMCAFF), a user posted that product managers around them are "queuing up to transform into FDEs," even viewing it as an antidote to "unemployment at 35."
The aforementioned transformer said that, combining his past experience and strengths, he was mainly considering consulting firms' specialized lines seconded to retail companies. He had actually already received an offer, but the company and compensation were not ideal enough. More importantly, since there are not many samples of this role, the performance evaluation system and long-term career prospects remain hard to see clearly.
"To put it bluntly, while it's a good transformation direction right now, I also worry that I'll gradually turn myself into an outsourced worker or a disposable employee. And if AI deployment gets fast enough, will this profession even be needed in the future?" he said.
In fact, in Chinese-language communities, many discussions about FDE include questions like "Old bottle, new wine—isn't this just domestic software outsourcing?" and "Domestic SaaS is finally blessed—haven't all the people doing on-site work for government and state-owned enterprises become FDEs?" There are also many small and obscure software companies with seemingly very traditional playbooks insisting that "what we do is FDE."
A headhunter currently recruiting for a senior FDE position explained the difference between this role and traditional on-site or implementation consultants: being able to use AI to write high-quality code, and having the ability and sense for AI solutions, is itself a new capability. More crucially, traditional on-site work mostly involves solving well-defined needs with fixed routines, whereas deploying AI faces mostly ambiguous pain points, requiring people who understand both business and AI to extract the real needs themselves and provide solutions on-site at the business. She especially emphasized the comprehensive strength of "combining business thinking, a consultant's mind, and an engineer's hands-on ability"—"one person who can act as an entire team."
The Fresh Research team analyzed that the FDE model is expected to reshape software delivery at scale, shifting from traditional license-based delivery to on-site co-building, while the relevant talent values cross-disciplinary capabilities spanning engineering, systems, product, and industry—and should not be confined to a single job category.
However, judging from the cases and phenomena reported by some media outlets and research institutions, "jack-of-all-trades" hexagonal talent is rare after all, and many of them have already advanced into executive positions. Meanwhile, the current salary levels of FDEs in China and their uncertain career prospects have, to some extent, weakened their appeal to senior talent. Some voices accordingly question that if the market supply consists mostly of "mediocre at everything" half-baked practitioners, the effectiveness of AI deployment will also suffer. In addition, many companies use FDE recruitment as a way to inflate the price of traditional outsourcing and on-site staffing arrangements, which has also somewhat diminished the value of the role.
As AI adoption accelerates, FDEs will likely remain in high demand. Clarifying the "gray areas"—defining what the role is for, what it can do, and what it should do—is crucial for AI companies, industry applications, and workers transitioning into FDE roles.