Inside the first batch of FDEs' survival stories: some landed ten-million-yuan deals, others were rejected at a 3,000-yuan quote

Before the National Day holiday, Shilong, a newcomer to the FDE role, submitted his resignation to the consulting company.

The thought had first crossed his mind in his very first week on the job. After a long period of agonizing, he finally took the step.

Although he was nearly capsized by this wave, he had no intention of leaving the FDE profession; he still wanted to hold tightly to the opportunities in it. Soon, he joined a second company.

Amid the boom in AI deployment, surging market demand has made FDE a red-hot field.

But when it comes to actual deployment, reality looks different: no one can say clearly how much AI deployment is really worth. Some have landed deals worth tens of millions, while others have only gotten enough money to buy WorkBuddy accounts and do data interface rework—the FDE delivery itself never commanded a real price.

"The demand is real, and so is the free-riding." Two months ago, Gao Feng started a business doing FDE. After two months of hustling, his biggest takeaway was: making money is exhausting, and the market is in complete chaos.

On one side, FDEs keep pouring in; on the other, a market with no standards has yet to form—the first batch of FDEs is struggling to survive in the chaos.

The chaos of the market is visible right from the quotes.

Some FDEs have already turned down four requests this year, because each was quoted at only 3,000 yuan, so they can only protect themselves by "declining orders." "It's not that I don't want to take them—it's that taking them wouldn't even cover the token costs."

But others are willing to chew on the hard bones, "even if it means losing money." To produce presentable cases as quickly as possible and land better orders later, they have to accept extremely unreasonable budgets.

A million-yuan project, shrunk to 100,000

Zhang Yang, a frontline salesperson, closed a deal with a 100,000-yuan budget over the past two months.

The client is a cross-border e-commerce trading company, and its needs and pain points are clear: replenishment forecasting—but this itself is no small challenge.

Order too little and you run out of stock; order too much and warehousing costs rise. And from production, export, customs clearance to shelving in local warehouses, the entire chain goes through a shipping cycle of more than 100 days.

What's more troublesome is that the client hasn't even sorted out its shipping data. Dozens of SKUs (stock-keeping units, such as products of different models and specifications) are shipped mixed together, and operations staff can only manually break them down and match them, confirming one by one how much of each SKU was shipped and whether it was all shipped.

Only then did Zhang Yang realize that the priority was to return to the most basic data governance and solve the problems of data matching and statistics.

After four or five rounds of talks, this AI forecasting project, originally expected to reach the million-yuan level, ultimately had to be downgraded. The deal the client settled on was 100,000 yuan, and that budget shrank to only enough to buy 10 WorkBuddy accounts plus data interface rework.

An 8 million quote, more than ten times the premium

He had heard that a domestic peer took on a supply chain FDE project for a multinational tech giant, quoting 8 million yuan. After learning the details, he found it even more "absurd": the project mainly involved data processing and then "asking a few numbers" so employees could directly query data such as sales and inventory.

Based on Liu Qiang's own project experience, this kind of work usually takes about two months, and the project amount generally ranges from 400,000 to 600,000 yuan.

The 8 million yuan quote was more than ten times higher than the project prices he knew of.

But he could also understand: the service provider might have to build capabilities from scratch, or the client's data foundation might be so poor that a lot of dirty, tedious work is required. But the key was still that "their boss has good connections, which is why they could sell it in."

In the AI deployment market, the quote for an FDE project is like Schrödinger's box: before opening it, it's hard to judge how much it is really worth.

After dealing with clients for a long time, Zheng Tao, who does FDE delivery, found that FDE projects are very different from the SaaS work of the past. SaaS is a standardized product with relatively clear pricing, but FDE delivery is often a solution that must be continuously adjusted according to the client's situation.

"There's a lot of subjective judgment in this; sometimes you can't explain the difference between 80% and 90%."

After experiencing the absurd reality of "a million-yuan project massively shrinking," Zhang Yang set two "rules": don't take projects from non-regular customers, and don't take projects that aren't requested by the "top boss."

The reason is simple: "It's hard to define delivery rules, and in the end it's hard to get acceptance."

The role dislocation of frontline FDEs

For FDEs just entering the field, the difficulties lie not only at the client site but also inside the company.

"Everyone is crossing the river by feeling the stones." Shilong, who entered the field full of hope, discovered after becoming an FDE that there is no ready-made path in this new profession, and that his leaders and seniors had no mature experience to pass on.

As his sense of professional value was worn away bit by bit, he fell into self-doubt: Am I really an FDE? What counts as an FDE?

Marginalized, reduced to a "freebie"

For this reason, he always felt that an FDE should at least first go to the frontline of the client's business—this role is closest to AI deployment, and its value lies in eliminating the translation costs of layer upon layer of relayed communication.

But as an FDE at this consulting company, he "wasn't qualified enough to contact clients directly and could only follow his leader," and he had never even been to a client site—he had practically become an outsourced worker. The client's meaning had to pass through consultants and leaders layer by layer, with information constantly degrading. "Sometimes even the leader himself may not be clear about what the client really wants."

No mature FDE working method had formed within the organization either; the division of labor was more like a traditional software team: one person sorts out business rules, another builds the system, and then they stitch it together through meetings and code merges—communication costs were extremely high.

Less than a month into the job, Shilong began to consider whether to stay or go. He could never find his place in the organization: he was neither at the client site nor able to participate in the company's core business.

The company he worked for mainly sold Huawei's methodology: it produced a proposal first, and only after the client nodded did it sign the contract and move in. The foundation of the business was the trust between consultants and clients. "They make money from this, not from an FDE's understanding of the business and then building out the coding."

The demos he made were merely tossed to clients for free, used to prove "we have the capability to land AI." He often mocked himself as a "freebie," or even "a freebie nobody wants."

After making up his mind to leave, Shilong soon received an FDE offer handed to him by a Feishu reseller, but he turned it down without much hesitation. In his view, that role was essentially still selling tools, which was different from his starting point for doing FDE. (For AI office vendors' FDE, is it selling tools or doing implementation? You're welcome to add the author on WeChat at skylar_12_14 to exchange frontline observations.)

Zhang Yang had struggled in the To B industry for many years and found that newcomers with little To B experience often underestimated the difficulty of landing AI.

Previously, a client wanted to build an Agent for market dynamics analysis, and a young person who had just entered the industry blurted out, "This is very simple, we've built one before too." Zhang Yang reminded him not to mistake a demo he could build for an AI system that could be deployed at an enterprise.

"The gap in between is enormous."

According to his experience, a demo built by an individual and an enterprise-grade system may look similar in function, but the workload and complexity are not on the same order of magnitude at all. He had once built a knowledge base with more than 20,000 lines of code, but to make it an enterprise-grade knowledge base, it might take 250,000 lines.

This kind of "veteran leading newcomers" setup was not uncommon in FDE teams.

Cui Jinyu, CEO of Zhigou Xinqing, was building an AI Native FDE team: an FDE project usually had two or three people assigned, with a senior business expert who understood AI more deeply serving as the pod leader, leading one or two juniors responsible for implementation and execution.

But this model placed great demands on how the organization worked. Under a traditional division-of-labor system, it was hard for new FDEs to have enough room to perform.

This was true of consulting firms, and software companies were no exception. Someone in the industry told Leifeng Net that most software companies claiming to do FDE still tended to first build an extremely complex SaaS middle platform, piling up a huge number of features, only for users to click in and find the same old stuff.

"Their people (FDEs) are trained this way too." Under the burden of history and institutional inertia, FDEs easily degenerated into on-site outsourced staff and were pushed to the margins.

Entrepreneurs' survival game

Unwilling to be bound by the old system, some chose to go it alone. For FDE entrepreneurship, the most important thing was "first let the company survive."

Entrepreneurs always have endless worries: at first they focus on acquiring customers and winning orders; after winning orders they have to solve conversion; once conversion runs through they precipitate methodology; when orders increase they get busy recruiting and expanding the team; later they find that B2B still requires building channels and an ecosystem.

But to survive, there was also a longer-term matter: how to prove you can stand your ground in an industry questioned by outsiders. To gain a foothold, you could not avoid games, compromises, or leveraging others' strength.

Free riding, impossible to guard against

Every time Gao Feng met a client, it was almost always a battle of wits: the other side might start by wanting both a proposal and training.

Getting enterprises to pay was not easy. The first lesson of FDE entrepreneurship was to face the reality of being free-ridden: after an hour-long conversation, all that was left was a polite remark. In the past two months, Gao Feng had been free-ridden no fewer than a dozen times.

At first, he did not know how to defend against it either. It required drawing an appropriate boundary—what counted as business investment and what counted as being free-ridden—but that was very hard to define.

Some requests seemed to clients like no more than a casual favor. But their questions came at him like machine-gun fire, and when he had answered them all, the other side left only one sentence: "Very professional, thank you, let's have dinner together next time."

Gao Feng often had to spend half a day or even a whole day dealing with these "casual" requests. Such a time cost was quite a luxury for an entrepreneur—when busy, he had to meet clients, job applicants, and partners in turn in a single day, and at his busiest he had to meet at least six groups of people.

Later he figured out some experience: first use twenty minutes to quickly go through the business requirements. If the other side was interested, let the business manager step in first, "let him block one round for me up front."

This was also the business wisdom many FDEs figured out after suffering losses: "Let people who purely do technology deal directly with clients, and sooner or later they'll be free-ridden to death." But rejecting clients could not be left to technical people. Once a technical person refused, the relationship might be hard to repair; if a salesperson did the refusing instead, even if the client got angry, the technical person still had a chance to salvage it.

But Gao Feng treated being free-ridden as tuition for figuring out this business.

"What I fear is the company being free-ridden on a large scale in the future." In his view, one person being free-ridden was not worth much money and could even accumulate some knowledge and experience; but once the company scaled up and everyone was free-ridden, the situation would be different.

New players, old rules

But whether one could survive was sometimes not something an individual or a team could decide.

Zhao Lei, head of another FDE startup, felt firsthand that the market was large and demand was rising, but competition was fierce, and a considerable portion of client resources was still held in the hands of established software companies.

What was harder was that traditional vendors still used practices such as man-day pricing in FDE projects, which easily framed FDE in the position of "advanced outsourcing." Once clients got used to this standard, it would be even harder for new players to prove themselves.

FDE entrepreneurs gradually realized that industry standards could not fall into others' hands; they had to seize the right to speak first.

In Cui Jinyu's view, FDE should not be a manpower service. "If it ends up becoming a software company or an outsourcing company, living exhaustingly and painfully, I'd rather go back to work at a big tech company."

What he envisions is an FDEaaS (FDE as a Software) model: letting AI gradually take over the work of traditional consulting and engineers—making proposals, writing code, and ultimately having these tasks fully taken over by Agents, with "people" retreating behind the scenes.

Some entrepreneurs have also begun to consider leveraging the power of channel partners: first helping channel partners build up their own FDE teams and shore up their delivery capabilities, then working together to open up customer resources.

The cost is that channel partners take a share of revenue, and the cut is not small, which erodes gross margins. But for startups, in the early days they can only accept it.

"After all, you have to survive first before you have pricing power." (How can FDE get its business model working? What successful experiences or cautionary tales are there? Feel free to add the author on WeChat at skylar_12_14 to discuss.)

Some change lanes, some force their way through

"Whoever can carve out a path from the chaos first has a better chance of surviving longer." This new profession of FDE is both hot and chaotic.

Some choose to change lanes and start over. Working as an FDE at a new company, Shilong has returned to the manufacturing industry he knows well, busy helping factories build a highly automated Agent system.

He shuttles daily between the company and client factories, living an "807" life—8 a.m. to 12 midnight, 7 days a week. But at last he is doing what he understands FDE should do, and he has found his place in the team: at least he is no longer a marginal "freebie."

Some insist on forcing their way through and refuse to leave the table. Cui Jinyu chooses to steady himself first. In his view, "although it can't be achieved overnight, you must plant a seed from the very beginning and wait for it to sprout."

But in a new profession, today cannot predict tomorrow. As Shilong often jokes about himself: "Do well and you're a pioneer; do poorly and you're a martyr."

Leiphone original. The author of this article focuses on big tech AI To B, and you are welcome to add the author on WeChat at skylar_12_14 to exchange and discuss.

Note: Gao Feng, Zhang Yang, Liu Qiang, Zheng Tao, and Zhao Lei in this article are all pseudonyms.