Frontier Advisory & Consulting Launches AI Agent "Frontier JSOX Agents" to Support J-SOX Compliance

【Overview】

Frontier Advisory & Consulting Co., Ltd. (Headquarters: Shinjuku-ku, Tokyo; President and CEO: Yohei Masada; hereinafter referred to as "the Company"), a professional firm specializing in the digital governance domain, is pleased to announce the launch of "Frontier JSOX Agents," an AI agent designed to support the practical aspects of internal control over financial reporting (J-SOX) compliance.

"Frontier JSOX Agents" is provided as a plugin that operates on Anthropic's "Claude" (Claude Cowork/Claude Code) used by client companies. It supports the repetitive J-SOX compliance tasks that occur each period by having the AI agent draft and check work with supporting evidence, while humans make the final judgments and confirmations. For example, by simply inputting existing RCMs (Risk Control Matrices), regulations, and evidence, it can draft everything from revisions to RCMs reflecting regulatory changes, to draft reports for operational status assessments through evidence matching, and to materials for responses to auditors.

The AI agent itself is free to use. Within the scope of the Company's "J-SOX AI Utilization Support Consulting," we provide support for customization tailored to the client company's controls, formats, and operations, as well as assistance in establishing its use. Even after the consulting contract ends, the agent can continue to be used free of charge.

【Background】

Maintenance of control documents such as RCMs, assessment of design and operational status, matching of evidence, aggregation of deficiencies, and responses to auditors – J-SOX compliance involves repetitive tasks each period, placing a significant burden on the personnel responsible. Manual work is prone to omissions and reliance on individual expertise, and requires considerable effort for explanations to auditors and organization of evidence.

On the other hand, these tasks involve a mix of "routine checks that should be performed without fail" and "areas requiring specialized judgment." We have developed this agent with the belief that by having the AI agent handle the drafting and checking of the former, and humans perform the final judgment and confirmation, we can improve both the response workload and quality.

【Features of "Frontier JSOX Agents"】

1. Role Sharing: "AI Agent Drafts, Humans Confirm"

This agent is not a tool for "AI to reach conclusions." By dividing roles between the AI agent and humans, with humans making the final judgments and confirmations, we achieve a quality that can withstand practical use and audits.

AI Agent: Standardizes terminology, automatically checks for completeness, consistency, and coverage, interprets evidence and documents, and drafts assessments and responses with supporting evidence and confidence levels.

Human: Approves, modifies, and confirms the AI agent's drafts. Specialized judgments, such as the severity of deficiencies or the necessity of disclosure, are always confirmed by humans and never automatically by the AI.

2. Auditability and Transparency

All processing is recorded in an audit log (a write-only record), providing evidence of "what the AI used as a basis for its draft and where humans approved, modified, or confirmed." This record can be verified when needed, and it can be confirmed that the record has not been altered (this is not a system for constant automatic monitoring and detection).

3. Data Remains Within the Client Company's Environment

This agent operates within the client company's Claude Cowork environment (Claude Code version is also available). Data is handled within the client company's environment, and there is no central database (it reads and writes files within connected folders).

【Scope of Support】

It supports the entire J-SOX compliance process, from scoping (selection of assessment scope) to maintenance and new creation of control documents, design and operational status assessments, progress management, aggregation and remediation of deficiencies, and responses to auditors. The controls covered are IT General Controls (ITGC), Process-Level Controls (PLC), and Entity-Level Controls (ELC).

Task

Capabilities

Determining Assessment Scope

Creation of materials for selecting key business locations, accounts, and target processes (top-down risk approach)

Maintenance of Control Documents

Creation of draft revisions and additions to RCMs reflecting changes in regulations, operations, and systems

New Creation of Control Documents

Creation of RCMs from scratch for new processes/systems (utilizing a list of standard risks and controls)

Inspection of Control Documents

Detection of inconsistencies between RCMs and regulations/procedures (broken references, insufficient coverage)

Design Assessment

Assessment of the appropriateness of control design and selection of key controls

Operational Assessment

Determination of population, sampling, evidence matching, identification of exceptions, and completion of documentation

Process-Level Controls (PLC)

Design and operational assessment linked to financial statement assertions (completeness check)

Entity-Level Controls (ELC)

Design and operational assessment in checklist/questionnaire format (six basic elements)

Preparation of Evidence

Collection of evidence, linking to controls/samples, detection of missing items, and creation of requests for acquisition

Deficiency Management

Aggregation of deficiencies, preliminary assessment of severity (confirmation by human), and follow-up on remediation status

Progress Management

Visualization of assessment progress across multiple controls and stages, reminders for pending tasks

Auditor Response

Creation of response materials for auditor inquiries and requests for information (PBC), and organization of submitted evidence

*The formats and operations described above can be customized to the client company's specifications through consulting.

【Service Delivery and Pricing】

Service Delivery: A plugin that operates within the client company's Claude Cowork environment (Claude Code version also available). Business users can operate it via chat, and specialized operational knowledge is not required.

Pricing: The agent itself is free to use. Consulting fees will be quoted individually based on the client company's situation and the scope of support.

Delivery Philosophy: The agent is provided as part of "J-SOX AI Utilization Support Consulting," which includes customization to the client company's specifications and support for operational adoption. Even after the consulting contract ends, the agent can be used free of charge (continued support is optional).

【Implementation Flow】

Implementation proceeds in stages through our consulting services.

Consulting Contract: Commence "J-SOX AI Utilization Support."

Customization and Adoption: Adjustments are made to the client company's specifications, and we provide support until operational adoption.

Full-Scale Operation and Ongoing Support: Transition to full-scale operation. The agent can be used free of charge even after the contract period (continued support is optional).

【Company Information】

Frontier Advisory & Consulting Co., Ltd.

Location: Shinjuku Nomura Building, 1-26-2 Nishi-Shinjuku, Shinjuku-ku, Tokyo Representative: Yohei Masada Business Activities: Consulting, digital solutions, and training programs primarily in the digital governance and internal audit fields, etc. Established: June 2020 URL: https://www.frontier-ac.com/

【Contact Information】

Frontier Advisory & Consulting Co., Ltd. Contact Person: Hideo Enari E-mail: [email protected]

URL: https://www.frontier-ac.com/contact