Article
What Is FDE Co-Development? The Success of New Businesses Depends on the "Number of Validation Iterations"
What determines the success or failure of a new business venture is not the precision of the initial proposal. It’s about launching the product into the market, gauging the response, making adjustments, and launching it again. The key lies in how many times you can repeat this cycle within a set timeframe. We believe that is the essence of the matter.
To use a baseball analogy, no matter how much you practice to improve your batting average, you won’t get more hits if you don’t step up to the plate often enough. From here on out, in this article, we’ll refer to a single attempt as a “at-bat” and the details of that attempt as a “verification cycle.”
And this “number of at-bats” is also the aspect of existing support services that is the hardest to sell. In this article, we’ll examine the structural reasons behind this and outline the path that led UPBOND to develop its “FDE Co-Development” service, using publicly available data.
If you have a project where testing has stalled, please contact us to discuss it.
What Is FDE Co-Development?
An FDE (Forward Deployed Engineer) is an engineer who works on-site at a client’s location, gains a deep understanding of their operations, and single-handedly handles everything from requirements definition to implementation and production deployment. UPBOND’s “FDE Co-Development” is a fixed-monthly-fee, co-development service that utilizes a small team consisting of this FDE (a person) and an AI engineer (monthly working hours allocated to an AI development agent), and commits to a specific number of verification cycles per quarter.
The reason this job role is rapidly gaining attention is that the final stage of new business initiatives falls outside anyone’s purview. A report by MIT Media Lab Project NANDA titled “The GenAI Divide” (2025) pointed out that the primary cause of failure in generative AI investments is not model quality, but rather the “last-mile” disconnect—where tools fail to adapt to on-the-ground operations. Even though a working prototype has been created, the process of deploying it into production remains a blank space.
It is not necessarily true that “having staff stationed on-site is too costly to be profitable.” Depending on how the structure is designed, it is possible to achieve both on-site staffing and high profitability.
The skills required for an FDE can be broadly categorized into three areas.
- Technical Skills: Implementation, Architecture, System Development
- Understanding Customers and Business Operations: Understanding on-site operations and translating challenges into requirements
- Problem Solving: Get out into the field, define the problem, and solve it completely
When the roles of defining requirements, building the product, and using it are separated, each handoff creates a need for verification and waiting time. FDE is a work style that consolidates these handoffs within a single person. UPBOND adds an AI engineer to this process, shifting tasks such as research, evaluation, prototyping, and documentation to the AI—thereby increasing the number of opportunities to take action without increasing costs.
Why Does the Number of Trials Determine the Outcome?
The success of a new business venture correlates more strongly with the number of validation cycles completed within a given period than with the accuracy of any single initial plan. This is because it is extremely rare for the very first plan to be a success right off the bat.
A successful project isn't one that was right from the start, but rather one where we were able to identify the mistakes early on.
Venture Studios are the organizations that have incorporated this premise into their systems. They operate a funnel process in which they evaluate 30 to 100 ideas and turn one of them into a company. The Global Startup Studio Network (GSSN) 2022 Data Report highlights the following differences:
However, depending on how the metrics are calculated, there is also data showing the opposite trend—namely, that traditional pre-seed VCs have a higher exit success rate—so comparisons of this kind depend on the calculation method.
On the other hand, the cost per at-bat continues to decline. Tomasz Tunguz of Theory Ventures has stated that “the cost of validating ideas has fallen by 95%” (though the basis for this calculation has not been disclosed), which aligns with reports that AI coding tools have reduced engineering effort by 20–35%. If at-bats have become cheaper, we can increase the number of at-bats.
Why Consulting Firms Don't Sell "At-Bats"
The reason the consulting industry does not base its pricing on the number of audits it performs is not a matter of ability or motivation, but because its revenue structure does not allow for it.
A typical team at a strategy firm consists of one partner, one engagement manager, and two to three analysts. This “junior-heavy” structure is known as “leverage,” and the following four factors work in tandem to support the profitability of the entire industry.
- Hourly Rate Variation: According to 2024 U.S. General Services Administration (GSA) procurement data, hourly rates are approximately $1,193 for senior partners, $834 for engagement managers, and $327–$498 for analyst associates. The range spans approximately 3.6 times.
- Fixed-Fee Agreements: The standard approach for high-value cases. The firm retains discretion over which attorneys to assign and for how many hours within that fixed amount.
- Project duration: Standard analysis projects last 8 to 16 weeks, while multi-phase transformation projects can last up to 18 months.
- The Economics of Utilization: While the target billable utilization rate is generally considered to be around 75%, the actual industry average is 66.4% (2025, based on industry benchmarks such as Kantata). To fill the gaps in utilization, projects are easier to manage when they are “long and concentrated blocks.”
In other words, the basic approach is to go all out on a single project, just once. Increasing the number of “at-bats” means breaking projects down into smaller pieces and shortening the cycle, which results in lower unit prices and shorter project durations. It’s a matter of not adopting that approach structurally —it’s not because we lack the ability.
There are exceptions, however. BCG Digital Ventures (which merged with BCG X in December 2023) has launched approximately 200 ventures with a success rate of 66% (two to three times that of traditional VC and CVC firms), and McKinsey Leap has announced plans to build more than 700 ventures by 2026; however, both are “offshore” entities operated on a separate accounting basis from their parent companies.
On top of that, AI-driven upheaval set in.McKinsey announced a reduction from approximately 45,000 employees in 2022 to about 40,000 by mid-2025, followed by a further cut of about 10% in December 2025. The cuts are concentrated in junior-level operational roles—such as research and report writing—areas where AI agents are increasingly taking over. The base of the pyramid is quietly beginning to hollow out.
Why You Don't Get a Chance to Step Up to the Plate at Major Japanese Corporations
The main reason why verification efforts are not progressing at large Japanese companies is not a lack of ideas. It lies in the time required for decision-making and in the fact that operations are launched without establishing evaluation criteria.
According to a survey comparing decision-making times, small and medium-sized enterprises (SMEs) take an average of 2 to 3 days to make strategic decisions, whereas large corporations reportedly take 28 days or more to make the same decisions. This difference stems from organizational hierarchy: large corporations have 8 to 12 layers of management between the front lines and senior management, while SMEs have 2 to 3 layers.
In fact, there is no shortage of hypotheses in the field. What’s stalled is the process leading up to the point where those hypotheses are put to the test. And even when they finally make it to the “batting box,” it’s not uncommon for the outcome to go unjudged. Because we rush into things without defining what constitutes success, no one can evaluate the results afterward. It has been repeatedly pointed out that the root cause of “PoC poverty” lies not in technology or methodology, but in a lack of strategic planning.
In the same survey, respondents cited “the lack of a system to evaluate new business creation processes, which discourages people from taking risks out of fear of failure” as a common challenge following the shortage of talent. From an institutional perspective, the single-fiscal-year budgeting system also poses a barrier. The practice of cutting the following year’s budget if funds are not fully utilized within the fiscal year hinders flexible allocation of funds.
Four Things UPBOND’s FDE Co-Development Service Promises
We chose not to offer recommendations or reports, but rather to make the verification process itself our product.
Consulting sells the "perfect swing." We've decided to sell approved at-bats by the number.
| What Consulting Firms Sell | What It Takes to Go from 0 to 1 |
|---|---|
| The Correct Swing | At-Bats × Learning Speed |
| Report | A Changing MVP and the Decision to Quit |
| Time × Leverage (Allocation is at the firm’s discretion) | Minimizing Cost Per At-Bat |
| Implementation is the customer's responsibility | Speed of Approval |
| Talent doesn't stay with the organization. | Assets That Grow Stronger With Every Investment |
1. We will commit to the number of verification cycles per quarter.
When we sign a contract, we don’t promise a specific number of man-hours; instead, we promise the number of verification cycles we’ll run within the specified timeframe. This shifts the basis for calculating costs from “how many person-months” to “how many times we can test.”
2. At the start of each cycle, we agree on the hypothesis, verification method, and decision criteria.
What do we verify during that cycle? How do we measure it? Under what conditions do we move forward, and under what conditions do we stop? We document and agree on these three points in writing before we begin. **We also count the decision to “stop” within the deadline as an achievement.** This is because being able to quickly confirm a failure creates an opportunity for the next attempt.
3. Situations requiring internal approval are also included in the scope of support.
It’s not uncommon for new business initiatives to stall not because of technical issues, but due to a lack of internal consensus. Our scope of work includes preparing proposal materials and quantifying business plans, and we translate the results of our analysis into a format that will secure approval. Shortening the “28 days” mentioned in the previous chapter is a task that even someone from outside the company can handle.
4. We will launch our Web3, wallet, and stablecoin initiatives on our own product platform.
We build our authentication, wallet, and payment systems on the product platform developed and operated by UPBOND itself. Since we don’t have to design the platform from scratch, we can get a head start, and as we gain more experience, shared components and insights accumulate within the platform.
Is your company’s planning process getting stuck before it “steps up to the plate”? We can help you assess your current situation and discuss your options — Contact Us
Rate Plans
The FDE Co-Development service is structured as a fixed monthly fee. The price is determined by the combination of FDE personnel and AI engineers.
Unit Price (excluding tax)
| Components | Unit Price |
|---|---|
| FDE (person) | 1.8 million yen per person-month |
| AI Engineer | 480,000 yen / person per month |
Standard Package (excluding tax)
| Plan | Monthly | Structure |
|---|---|---|
| Trial | 600,000 yen / month | FDE 0.2 + 0.5 AI Engineers |
| Standard (Recommended) | 1.38 million yen / month | FDE 0.5 + 1 AI Engineer |
| Full | 2.76 million yen / month | FDE 1.0 + 2 AI Engineers |
Adjustments to fit your budget will be made not through discounts, but by modifying the FDE utilization rate and the composition of AI engineers. We keep the unit price fixed so that you can compare, across projects, how many work shifts can be covered for the same amount of money.
Frequently Asked Questions
What is the difference between FDE co-development and SES/contract development?
SES guarantees operating hours, while contract development guarantees the completion of specified deliverables. What FDE’s collaborative development promises is the number of verification cycles to be run within the designated timeframe and the decision (proceed or halt) made at the end of each cycle. Since the same team handles everything from requirements definition through implementation and production deployment, there is no back-and-forth between the party providing the specifications and the party building the system.
What exactly do you do during a single verification cycle?
At the start of the cycle, we will document and agree on three key elements: the hypothesis to be tested, the verification method, and the decision criteria. Based on this, we will build a working model (prototype or MVP), test it with actual users and in real-world business operations to measure the results, and then use the agreed-upon criteria to decide whether to proceed or discontinue the project. We will summarize the results of this decision and provide them in a format suitable for internal approval.
Can this be used in conjunction with an existing consulting contract?
They can be used together. FDE co-development is not intended to replace strategic recommendations or market research; rather, it handles the process of actually bringing those recommendations to market to test them. If you already have a concept or business plan, we will design the first cycle based on that.
What is the minimum contract term, and what happens if the contract is terminated early?
Since we commit to a specific number of validation cycles on a quarterly basis, our standard contracts are structured in three-month increments. If you’d like to start small, you can begin with the Trial Plan (600,000 yen per month, excluding tax). Specific terms regarding duration, renewal, and termination will be outlined in an individual contract based on the details of each project.
Summary
- The success or failure of a new business depends not on the precision of the business plan, but on the number of validation cycles completed within the given timeframe. 92% of startups pivot before reaching product-market fit.
- The reason consulting firms don't sell "at-bats" isn't a matter of ability; it's a consequence of their revenue structure—a fixed fee multiplied by leverage. The industry average for billable utilization is 66.4%.
- The reasons why initiatives stall at major Japanese corporations are decision-making processes that take 28 days and Proof of Concept (PoC) projects that begin without establishing criteria (approximately 70% lack KPIs).
- **FDE (Forward Deployed Engineer)** is a role in which a single person handles everything from requirements definition to implementation and production deployment, eliminating the wait times that arise at the boundaries between processes.
- UPBOND's FDE co-development service involves a small team of FDE and AI engineers who commit to a fixed monthly fee based on the number of verification cycles conducted per quarter.
It’s not about how precise the plan needs to be, but rather when you can release the first version. Simply increasing the number of “at-bats” from one or two per year to four will make a difference in the amount of learning you retain for the following year.
If you have a project where verification has stalled, or if you are interested in FDE’s collaborative development services, please contact us to discuss your needs.