Tax accountant with no programming experience develops business app in "half a day" - if Inc. releases case study of tax accountant firm internalizing bookkeeping outsourcing using Claude Code

if Inc. (Nakano-ku, Tokyo / Representative Director: Yasuyuki Ito) announced on July 7, 2026, a case study of its "if AI Partner" generative AI total support service. The case study details how a sole practitioner tax accountant firm, after implementing "Claude Code," an AI for writing code, has reduced its business app development time from approximately one week to half a day (self-reported, perceived time). The firm is now in the process of putting its internally developed journal entry app into practical use and transitioning bookkeeping tasks, previously outsourced, to in-house processing. "if AI Partner" is a generative AI total support service provided by if Inc. for management, DX promotion, and HR personnel at small to medium-sized enterprises. It primarily offers individual support plans with weekly consultations, can be combined with corporate training, and supports the utilization of subsidies such as the Human Resources Development Subsidy and IT Introduction Subsidy. The key point of this case study is that a tax accountant with no prior programming experience has moved beyond the stage of "being able to create apps" to using the created apps to replace outsourcing, have staff use them in daily operations, and is now contemplating offering them as client services.

Case Study Overview | Sole Practitioner Tax Accountant Firm x Claude Code Short-Term Development Prerequisites

Industry: Sole practitioner tax accountant firm (firm name and individual name withheld for anonymous disclosure)

Scale: 1 full-time employee / approx. 50 regular clients / approx. 30 sole proprietors filing tax returns

Prerequisites before implementation: No programming experience / No Excel macro experience

Support Service: if AI Partner (individual support plan, weekly consultations). Gradual support from building a foundation for generative AI utilization, transitioning to business app development with Claude Code.

Recent Key Achievements: 1 Business app development time reduced from approx. 1 week to half a day (self-reported, perceived time) / 2 Journal entry app operational in actual business and staff usage commenced / 3 Transitioning previously outsourced bookkeeping tasks to in-house processing (in progress) / 4 Commencement of planning for client service offerings.

Disclosure Policy: All figures are self-reported and perceived values of the individual, not performance indicators verified by a third party.

Background: From building a foundation for generative AI utilization to implementing Claude Code - the phased approach of if AI Partner's support

The management of the firm in this case study initially began by utilizing generative AI in their business operations (automating data input pipelines, differentiating tool usage). They first built a foundation by improving the efficiency of existing tasks, such as reducing the time for data input from bank statements and transaction details (500-600 lines) from about 2 days to about 30 minutes (self-reported, perceived time). Subsequently, they introduced "Claude Code," an AI for writing code, through a phased learning process supported by weekly individual consultations from if AI Partner.

Immediately after implementing Claude Code, they developed their first web application (a client document submission page) in approximately one week (16-20 hours of actual work, perceived time). From here, the development speed began to accelerate.

Recent Achievements After Implementing Claude Code: Development time "approx. 1 week -> half a day" - Journal entry app in practical operation - Bookkeeping outsourcing to be internalized

The changes confirmed during weekly consultations over the past 1-2 months are as follows (all self-reported, perceived time):

Business app development in "half a day": In response to a request from a real estate management company client, they developed an app for real estate owners to create remittance statements. This app reads PDF bank statements, identifies the source of deposits and properties, and generates PDF remittance statements. The development took half a day. As a tax accountant familiar with back-office operations, they were able to design the requirements themselves and propose the solution without detailed consultations, stating, "I thought this is what should be made." During development, they delegated tasks to the AI and worked in parallel with other duties.

Journal entry app operational in actual business: Their internally developed journal entry app, which reads PDFs of receipts etc. using OCR, determines account categories, and outputs in a format importable by accounting software, has started operating in actual bookkeeping tasks. The execution environment has also been introduced to staff, transitioning to a system where each person uses the app themselves.

Transitioning bookkeeping outsourcing to in-house: Previously, a portion of bookkeeping tasks was outsourced, taking about 10 business days for processing, with costs reaching approximately 200,000 yen during peak periods (self-reported). With the operation of their in-house app, they are gradually switching previously outsourced bookkeeping tasks to in-house processing. Due to improved processing speed, they anticipate being able to "get ahead on work during the summer lull."

Operational design that "won't break even if AI advances": OCR misjudgments are stored in a database as rules, and the system is designed to re-evaluate and regenerate CSV output (OCR itself is not redone). The execution part uses local, self-contained OCR, making it AI-independent. Learning is concentrated on the rule database side, achieving operations that are less susceptible to changes in external service specifications. Additionally, a function to check the consistency between the total amount from OCR results and manually entered values ensures accounting quality.

Security and Governance for Professional Services: A security plugin officially provided by the developer for checking the vulnerability of developed items has been introduced. As they handle client financial data, they have decided to transition to a corporate plan and have confirmed the contract terms regarding data processing. They also plan to add a memorandum regarding AI and cloud usage to existing client contracts.

*The above content is based entirely on the self-reporting and perceived experiences of the individuals involved in this case study and are not performance indicators verified by a third party. It does not guarantee that similar results will be achieved with similar initiatives.

Comments from Case Study Participants (Management of Sole Practitioner Tax Accountant Firm - Summary of Statements)

"The first app took about a week, but now I can develop one in half a day. As a tax accountant, I know back-office operations, so I can think of the requirements myself and propose them without detailed consultations, thinking, 'This is what needs to be made.' During development, I leave it to the AI and only occasionally check the screen while talking to employees on the phone. More than the act of creating itself, I feel it's significant that we are regaining outsourced bookkeeping tasks with the apps we've created, that our staff are using them, and that we're starting to see possibilities for offering them to clients."

*This is a summary of statements based on weekly consultation records, edited for readability (anonymously disclosed).

Comments from Support Staff (Fumitaka Toyomatsu, Generative AI Department, if Inc. - Summary of Statements)

"When we started our support, we were at the stage of checking tool usage one by one together. Now, before we even ask for a hearing, the individual designs the requirements and shapes an app in half a day. I feel this is significant progress. This case study confirmed that a phased approach, starting with building a foundation for generative AI utilization before moving to Claude Code, is effective for in-house development by professionals."

What is if AI Partner: Comprehensive support from building a foundation for generative AI utilization to in-house development of Claude Code

Service Name: if AI Partner (Generative AI Total Support)

Provided by: if Inc.

Target Audience: Management, DX promotion personnel, HR personnel of small to medium-sized enterprises, etc. (including small firms such as professional service providers).

Format: Individual support plan (primarily weekly consultations) / Can be combined with corporate training.

Scope of Support: AI strategy planning / Business process redesign / AI tool selection and appropriate usage / Support for in-house development of business apps using Claude Code, etc. / Clarification of security and governance aspects / Support for subsidy application procedures.

Subsidies: Supports the utilization of the Human Resources Development Subsidy and IT Introduction Subsidy (as of July 2026. Please contact us for the latest system information).

*Subsidy utilization is subject to prescribed requirements and screening. Subsidies are not guaranteed. Please contact us for details.

if AI Partner Implementation and Application Flow

Free Implementation Consultation (30 minutes, online): We will listen to your current operations and challenges.

Plan Proposal: We will propose a plan after organizing the consultation frequency, scope of support, and eligibility for subsidies.

Start of Support: Primarily through weekly consultations, we will provide phased support from building a foundation for generative AI utilization to in-house development.

Service Details (LP): https://if-tech.co.jp/ai-partner

Inquiries: [email protected]

Frequently Asked Questions (FAQ) | Regarding Claude Code Utilization and Tax Accountant Firm In-house Development Case Study

Q. What are the figures like "development in half a day" and "approx. 1 week -> half a day" based on? A. All figures are based on the self-reporting and perceived experiences of the individuals involved. They are not performance indicators verified by a third party, nor do they guarantee equivalent figures for similar initiatives.

Q. Will the firm name and manager's name in the case study be disclosed? A. This press release is an anonymous disclosure, with the firm name and individual name withheld. Only the industry, scale, and details of the initiative are disclosed.

Q. Can someone with no programming experience really develop a business app in half a day? A. The management of the firm in this case study reports that after building a foundation for generative AI utilization through phased support, they can now develop business apps in half a day, having previously taken about a week immediately after implementing Claude Code. There are individual differences in prerequisites and time required, and equivalent results are not guaranteed.

Q. Has bookkeeping outsourcing been completely stopped? A. Due to the operation of the internally developed journal entry app, they are in the process of switching previously outsourced bookkeeping tasks to in-house processing (as of July 2026, self-reported).

Q. How can I apply for an if AI Partner consultation? A. We are accepting 30-minute free implementation consultations. Please apply through the service site above.

Future Developments: Towards "AI Back Office" for Professional Services and Offering Client Services

The firm in this case study is not only improving its own operations but is also considering plans to offer developed apps to clients at low prices (e.g., subscription service for the remittance statement app for real estate management companies) and to develop new "AI Back Office" services by combining internal control knowledge with app development. They are also working on a strategy to commonize parts such as payroll calculation and invoice processing and customize them for each client, as well as automating the input of large volumes of receipts.

if Inc. will continue to release case studies of in-house development and service offerings by professionals through if AI Partner.

About if Inc.

if Inc. (If) operates under the vision "To a world where anyone can create with technology" and provides IT human resource development, generative AI human resource development, and corporate reskilling training through its "Tech Mentor" brand. Beyond training, they provide comprehensive support for companies' generative AI utilization, from strategy planning to construction, education, and implementation.

Company Name: if Inc.

Representative: Representative Director Yasuyuki Ito

Established: January 2021

Location: 2-17-1 Nogata, Nakano-ku, Tokyo

Email: [email protected]

Company Website: https://if-tech.co.jp/

Business Activities: Provision of "if AI Partner" generative AI total support service / Provision of "Tech Mentor Corporate Training" for corporate reskilling / Operation of programming school "Tech Mentor" / Contract development and provision of AI-driven teams.

<Contact for Media Inquiries> if Inc. Public Relations / E-mail: [email protected]

*The firm name and individual names in the case study described in this release are withheld. The figures for development period, costs, business hours, etc., described are based on the self-reporting and perceived experiences of the individuals involved and the situation as of July 2026. They are not performance indicators verified by a third party, nor do they guarantee that similar results will be achieved with similar initiatives. Product names (Claude Code/Claude, etc.) are cited for reference to generally available tools, and this release does not recommend or comparatively evaluate each product.