The Untapped Gold Mine: Building a Digital Business Succession Service with Claude Code and Codex

Digital Business Succession Rescue: Turning Japan's "Personalized Systems" into Transferable Assets

"What can I build with Claude Code and Codex to make money?"

When faced with this question, most people immediately think of contract AI app development, business automation, chatbots, Japanese versions of overseas services, or small SaaS products.

However, those areas are already overcrowded.

As the speed of writing code with AI increases, the value of simply being able to "write code" decreases.

From a client's perspective, Company A, B, and C all say, "We use AI for short-term development." Eventually, this leads to price competition, where the time saved by AI is pushed back as pressure to lower prices.

So, what is truly valuable in the AI era?

The answer is not the code itself.

It is the act of demystifying business processes that no one can explain and creating a state where the company doesn't stop even if someone leaves.

The strongest gold mine to target by combining Claude Code and Codex is the following business:

"Digital Business Succession Rescue": Discovering business rules buried in Excel, VBA, Access, legacy systems, manual tasks, and the heads of employees, and converting them into specifications, tests, migration plans, and operational procedures.

This is not mere system development.

It is not simple DX (Digital Transformation) consulting either.

It is the job of rescuing "invisible software" buried within a company at critical moments such as a president's retirement, a veteran employee's departure, an M&A transition, or the failure of an old PC.

What you are selling is not a program.

It is the peace of mind that "the company will keep running even if that person is gone tomorrow."

1. Why This is a Massive Gold Mine in Japan

There are approximately 3.36 million small and medium-sized enterprises (SMEs) in Japan.

According to the Small and Medium Enterprise Agency, as of June 2021, SMEs and small-scale businesses account for 3.365 million entities, supporting roughly 70% of Japan's employment.

In other words, the bulk of the Japanese economy is powered not by the latest systems of large corporations, but by the daily operations of local SMEs.

Source: Small and Medium Enterprise Agency "Number of SMEs and Small Businesses"

On the other hand, these sites are filled with systems like:

• Excel macros created by employees who have already retired
• Access databases that only one administrative staff member knows how to operate
• Sales management systems that have been patched for 20 years
• Tasks involving manual CSV processing for accounting software
• Order processing combining email attachments, shared folders, and paper checklists
• Exceptional handling for discounts, closing dates, inventory allocation, and billing known only to specific staff
• Batch files and old scripts that no one wants to touch
• Databases whose server location is unknown

Crucially, these are not just "old systems." They contain decades of accumulated business know-how regarding transaction terms, customer service, inventory judgment, quality control, and billing.

Even if the code is messy or the screens are outdated, if it supports monthly sales and payments, it is a vital asset for the company.

However, in many cases, these assets are not documented in specifications or manuals. They exist fragmentedly in the heads of employees, Excel formulas, VBA branches, Access queries, and shared folder filenames.

This is where the business succession problem overlaps.

The 2025 White Paper on Small and Medium Enterprises shows that more than half of SME owners are over 60, and about 40% of sole proprietors intend to discontinue their businesses. While the rate of missing successors is declining, aging management and business succession remain major challenges.

Source: Small and Medium Enterprise Agency "2025 White Paper on SMEs"

Even if shares or real estate are handed over, if the actual operations from order to billing are only understood by one person, the company hasn't truly been handed over.

The same applies to M&A. If it's discovered after the acquisition that "we can't invoice without this Excel file" or "inventory counts are in a staff member's memo, not the system," the integration stops.

In short, the business succession happening in Japan requires "Digital Business Succession"—the transfer of business rules, data, systems, and procedures.

2. The Market Lacks People, So AI Leverage Works

Trying to do this with traditional labor-intensive methods is not profitable. You would need people who can read old code, interview staff, analyze data, design new systems, and write tests.

However, the shortage of DX talent in Japanese companies is severe. According to an IPA survey, 85.1% of Japanese companies reported a shortage of talent to promote DX.

Particularly in small businesses, while there is interest in generative AI, the actual utilization rate is low. Only about 20% of companies with fewer than 100 employees are actively using generative AI for work.

Source: IPA "DX Trends 2025" Materials

At first glance, this looks bad for AI businesses. In reality, it's the opposite.

The client doesn't need to operate Claude Code or Codex. The provider uses the AI to shorten the hundreds of hours usually required for investigation, code reading, spec extraction, test creation, and data comparison.

You aren't selling AI to the client. You are selling:

• Reduced risk of business stoppage
• Completion of handovers from retiring staff
• Integrability after M&A
• Recoverability from system failures
• Readable business rules
• Verifiable tests
• Criteria for what to abolish and what to keep

The less the client can use generative AI, the higher the value of the provider who uses it behind the scenes to deliver results.

This is key. In "Generative AI Implementation Support," the client must learn new ways of working. But in "Digital Business Succession Rescue," the client doesn't have to learn anything. They just explain their work, provide files, and confirm the final rules. You sell results, not AI.

3. Why This Wins Over Other AI Business Ideas

Let's compare business ideas using Claude Code and Codex across five axes (1-5 scale):

1. Customer Urgency
2. Ability to Pay
3. Productization/Repeatability
4. Cost Reduction via AI
5. Long-term Entry Barriers

General AI App Contract Development

• Urgency: 2 | Unit Price: 3 | Repeatability: 3 | AI Leverage: 4 | Barriers: 2 Budget is hard to secure when clients just think it "would be nice." It often leads to price wars.

Prompt Training/AI Courses

• Urgency: 2 | Unit Price: 2 | Repeatability: 3 | AI Leverage: 3 | Barriers: 1 Easy to start, but competition is high and content becomes obsolete quickly.

Small Industry-Specific SaaS

• Urgency: 3 | Unit Price: 3 | Repeatability: 4 | AI Leverage: 4 | Barriers: 3 High potential for recurring revenue, but requires high initial accuracy in identifying industry pain points and heavy support/sales.

Japan Expansion of Overseas AI Services

• Urgency: 3 | Unit Price: 3 | Repeatability: 3 | AI Leverage: 4 | Barriers: 2 First-mover advantage exists, but disappears if the original service adds Japanese support.

Digital Business Succession Rescue

• Urgency: 5 | Unit Price: 4 | Repeatability: 4 | AI Leverage: 5 | Barriers: 4 This business has clear deadlines: the day a staff member retires, the day the president steps down, the M&A execution date, or the end of server support. Urgent problems get budgets. Furthermore, the entry point is diagnosis, but the exit is long-term infrastructure rebuilding.

4. The True Meaning of "No One Has Noticed"

To be precise, the market for legacy system modernization already exists for large enterprises, served by major SIers and consulting firms. The "empty space" is here:

Projects too small for majors to take, but too complex for local IT shops to investigate manually.

For example, a company with 20 to 200 employees using dozens of Excels, Access files, and old sales systems. The project size is a few million yen. Traditionally, the investigation alone would eat the profit. But if you use Claude Code and Codex to automate investigation, documentation, and testing, you can offer these as fixed-price products.

This isn't inventing a new market; it's compressing modernization that only worked for big companies into a product for SMEs.

5. Why Use "Both" Claude Code and Codex?

Claude Code and Codex are not just code completion tools; they are agentic environments. Claude Code can read codebases, edit files, and run commands. Codex also handles development, refactoring, and migration with non-interactive execution and Git integration.

You use both not to find a winner, but to separate the "creator" from the "skeptic."

If one model does everything from extraction to testing, it might carry its initial misunderstandings through the whole process. Instead, divide roles:

Claude Code: The Archaeologist & Architect

Investigates existing assets. Reads code, SQL, logs, and manuals to organize dependencies and business rules. Sub-agents can handle data structures, business rules, and manuals separately.

Codex: The Factory & Independent Verifier

Reads the specs from Claude and verifies if they match actual code behavior. It then runs parallel tasks like test creation and data conversion. It uses CI to check for exceptions.

Human: The Final Business Judge

Even if models agree, humans approve. AI should not finalize decisions on whether an exception is still needed or if tax processing is correct.

6. The Product is an 8-Step Rescue Process

Don't start by building a new system. First, document what is running and decide what to keep. The process is productized into 8 stages:

1. Preserve Evidence: Record backups, file lists, and server configs. Don't modify the production environment yet.
2. Inventory Digital Assets: Collect everything—VBA, scripts, CSVs, paper checklists, and staff memos.
3. Record Actual I/O: Capture typical input data and resulting outputs (e.g., normal orders, discounts, returns) to see how the code actually behaves.
4. Map Business Rules: Convert code into structured business rules (Input, Condition, Formula, Priority) that management can read.
5. Create Golden Master Tests: Fix current inputs and outputs as a baseline to ensure new processes produce the same results.
6. Classify (Abolish, Replace, Envelop, Rebuild): Don't rebuild everything. Use SaaS where possible and only rebuild the core competitive logic.
7. Incremental Migration: Use Codex worktrees to migrate units (e.g., order intake, price calculation) one by one, running old and new in parallel.
8. Deliver the Succession Pack: Hand over a pack containing asset lists, flowcharts, rule ledgers, recovery procedures, and a roadmap.

7. Example: Rescuing a "Mystery Excel" at a Local Wholesaler

A wholesaler with 45 employees relies on a 10,000-line VBA macro for price calculations. The creator is gone; only one 59-year-old knows how to run it. Instead of a full system replacement, the Rescue service extracts the rules, finds that 8 of 20 processes are unused, 7 can be handled by SaaS, and only 5 are unique. You rebuild just those 5. The value is the completed handover and documented rules, not just the code.

8. Product Design and Pricing

• Product 1: Black Box Risk Diagnosis (5-10 days, 300k-800k JPY): Delivers an asset list, identification of person-dependent tasks, and an improvement plan.
• Product 2: Digital Succession Pack (4-8 weeks, 1.2M-3M JPY): Establishes business rules, data dictionaries, and recovery procedures.
• Product 3: Incremental Migration Sprints (2-6+ months, 3M-15M+ JPY): Migrates units to SaaS or APIs.
• Product 4: Continuous Maintenance (100k-500k JPY/month): Monthly backups, test execution, and handover updates.

9. Revenue Model for a Solo Operation

A solo operator could earn over 20 million JPY annually by securing 12 diagnoses and 4 Succession Packs, plus a few monthly maintenance contracts. The market is huge; even if only 0.05% of Japan's 3.36M SMEs buy a 2M JPY service, it's a 3.3 billion JPY market.

10. Target Industries

Focus on companies with 20-200 employees in wholesale, manufacturing, or logistics with complex order/inventory tasks. Specifically target back-office tasks like "Order to Billing," "Inventory and Ordering," and "Monthly/Yearly Closing."

11. Sales Strategy: Use "Trusted Intermediaries"

Don't sell directly to CEOs via ads. Partner with those who already have trust: tax accountants, regional banks, M&A brokers, and local IT maintenance firms. Use questions like: "Which task stops if one person is absent tomorrow?"

12. The Real Entry Barrier

It's not the AI tools. It's the accumulated assets:

1. Business rule dictionaries for Japanese commerce.
2. Analysis templates for legacy assets (VBA, Access, etc.).
3. Interview designs to extract "hidden" rules.
4. Golden master test components.
5. Industry-specific migration patterns.

13. What NOT to Leave to AI

Never give AI write access to production data initially. Separate original files from working copies. Do not hardcode secrets. Most importantly, AI explanations are not evidence—link them to actual code and I/O results. Final legal/tax/accounting decisions must be made by experts.

14. 90-Day Execution Roadmap

• Days 1-14: Pick one industry and build a demo environment with fake legacy files.
• Days 15-30: Interview 10 intermediaries (tax accountants, etc.) to refine the product.
• Days 31-60: Perform 2 paid diagnoses at a low price to verify the process.
• Days 61-90: Complete one Succession Pack and turn it into a partner-ready product.

15. The Ultimate Goal: A Digital Succession Platform

Eventually, evolve from a service to a platform featuring automated inventory agents, business rule ledgers, "Succession Scores," and a "Business Succession Room" for M&A buyers to verify risks.

Conclusion: Don't Sell Code. Sell "Business Continuity."

In the AI era, the cost of writing code drops, but the cost of understanding "what to build" and "how it currently works" remains high. Claude Code and Codex are the pickaxe and ore separator. The real product is the assurance that the company will keep running tomorrow, even if its key person leaves.