Terms of Service (Catalog Specification Extraction)

Article 1 (Scope)

1. These Terms govern the use of the cloud service "Zag Catalog Specification Extraction" (the "Service") by any company or organisation (the "Customer") and the individuals it authorises (the "Users").

2. Where Pioneerwork and the Customer have executed a separate agreement (service agreement, SLA, NDA, data processing agreement — an "Agreement"), the Agreement prevails over these Terms.

3. Trial use with an access code issued by Pioneerwork is also governed by these Terms; trial-specific conditions are in Article 8.

Article 2 (Definitions)

"Input Data" means documents the Customer registers (PDF, spreadsheets, CSV, images, documents fetched by URL), data supplied from the Customer's product information systems, and edits Users make on screen. "Output Data" means specification tables, attribute definitions (schemas), extracted values, evidence (page and position), export files and the results reflecting User edits generated by the Service. "AI Functions" means the parts of the Service that use large language models or other machine-learning models to read, classify, extract and propose. "AI Provider" means the third party listed in Annex 1 whose platform runs the AI Functions.

Article 3 (Changes)

Pioneerwork may amend these Terms with at least 14 days' notice by posting on the Service or its website or by e-mail; material changes adverse to the Customer require the Customer's consent under the Agreement.

Article 4 (Notices)

Notices to the Customer are given by e-mail to the registered contact or on the Service; notices to Pioneerwork go to the contact on www.withzag.com or the contact named in the Agreement.

Article 5 (Access codes)

1. The Customer must keep access codes confidential and disclose them only to authorised Users. Use under an access code is deemed use by the Customer unless the leak is attributable to Pioneerwork.

2. On suspected compromise the Customer notifies Pioneerwork, which promptly revokes and reissues the code.

Article 6 (Roles)

1. In production, Pioneerwork provides a login portal and an administration portal where the Customer's administrators create departments and Users and assign rights (view, edit, approve, export). Granularity (per department, per product project) and approval flows are defined in the Agreement.

2. Pioneerwork's operations staff access the Customer's environment with administrator rights only for operation, maintenance and support; every such access is recorded in the audit log.

Article 7 (Environment)

The Customer provides its own devices, network and browser (current Google Chrome or Microsoft Edge are verified). Pioneerwork advises on connectivity through the Customer's proxies or security gateways; configuring those devices is the Customer's responsibility.

Article 8 (Trial)

1. Trials evaluate extraction accuracy, schema granularity and usability, free of charge, for the period Pioneerwork notifies.

2. The trial runs on the same system and extraction engine as production, but without user management, role settings, SLA coverage and any functions Pioneerwork expressly excludes.

3. Trial access codes carry a shared user role, not administrator rights.

4. Trial data is handled under Article 22 and can be carried into production on the Customer's instruction.

5. Trials are provided without warranty of any kind.

Article 9 (Functions)

1. Document registration (PDF, spreadsheets, CSV, images; several files per job).

2. Reading and structuring: page triage, reading of specification tables, dimension drawings and feature pages, schema generation, per-model value extraction, product classification against the taxonomy the Customer supplies.

3. Evidence: page, position and printed text recorded per value and shown on screen.

4. Review and editing: value states (read, needs review, not printed), User edits, column merge/split/add with AI suggestions, annotations.

5. Export: specification tables (CSV etc.) and schemas.

6. Administration: job list, processing status, audit log (production).

Article 10 (Annotation by Pioneerwork)

1. To assure quality, Pioneerwork staff and AI may review a job's results, correct values found to be wrong, fill values that were missed and add evidence ("Pioneerwork annotations").

2. Every annotation is audit-logged and shown as "Annotated by ZAG"; prior values are retained and the Customer may request reversal at any time.

3. Annotations rest on what is printed in the Input Data; no value is ever guessed.

Article 11 (Changes, suspension, termination)

1. Pioneerwork may add or change functions; changes with material impact are announced in advance.

2. The Service may be suspended for maintenance, incidents, platform outages or force majeure; planned maintenance is announced in advance and scheduled outside the Customer's business hours where possible.

3. Discontinuation of the Service requires at least 90 days' notice and return/deletion of data under Article 22.

Article 12 (Support)

Pioneerwork answers questions on operation, defects and extraction results through the channel and hours set in the Agreement. Wrong results are corrected on the Service after verification against the evidence, with the correction recorded.

Article 13 (Nature of AI output)

1. AI output is probabilistic and may contain errors, omissions or misreadings.

2. The Service is designed so that errors surface and are corrected: evidence matching against the print, visible review states, Pioneerwork annotations and User edits.

3. The Customer verifies Output Data to the extent needed before relying on it.

Article 14 (AI Provider)

1. AI Functions run on Microsoft Azure OpenAI Service in the Japan region (Annex 1). Changes of provider or region are notified in advance.

2. Transmission to the AI Provider is governed by Microsoft's enterprise terms (Product Terms and Data Protection Addendum).

Article 15 (No training)

1. Input Data and Output Data are never used to train or improve AI models of Pioneerwork or the AI Provider.

2. Under Microsoft's terms, prompts and completions sent to Azure OpenAI Service are not provided to OpenAI, are not available to other customers and are not used to train or improve Microsoft or third-party models.

3. Pioneerwork may use statistics that cannot identify any document (counts of review states, processing times) to improve the Service; document content is never used for model training.

Article 16 (What is sent to the AI)

1. Transmission is limited to the minimum each step needs.

2. Sent: the image and text of the page being read (specification tables, dimension drawings, feature pages), the table structure extracted from that page, attribute names (schema) defined by Pioneerwork or the Customer, and the names of the product categories used for classification.

3. Not sent: account information (access codes, user names, e-mail addresses), data of other jobs, metadata such as file names, uploader or timestamps, and the Customer's internal data supplied separately (e.g. product information system exports), except where the Customer explicitly instructs their integration into a table.

4. Calls are made per page or per table; a whole document is never sent at once.

Article 17 (Retention at the AI Provider)

1. Under Microsoft's terms, Azure OpenAI Service may store prompts and completions, encrypted, for up to 30 days solely for abuse monitoring; such data is not used for training and is accessible only to authorised Microsoft personnel investigating abuse.

2. If the Customer requires exemption from that retention, Pioneerwork applies to Microsoft for it and reports the outcome.

3. Pioneerwork records AI requests and responses as part of the job data for quality control and incident analysis; those records are handled under Chapter 5.

Article 18 (Input Data)

Rights in Input Data remain with the Customer or its licensors. The Customer grants Pioneerwork the right to copy and process Input Data only as needed to provide the Service (reading, storage, display, annotation, export, backup, support) and warrants it is entitled to register the data.

Article 19 (Output Data)

1. Rights in Output Data (ownership, copyright and other intellectual property, to the extent they arise by law) belong to the Customer. Pioneerwork claims no rights beyond what providing and supporting the Service requires.

2. The Service software, extraction methods, screen designs, generic attribute taxonomies and their improvements remain Pioneerwork's under Article 31.

Article 20 (Location and separation)

1. Input and Output Data are stored in Microsoft Azure, Japan East region, operated by Pioneerwork.

2. Data is logically separated per customer. A dedicated environment (own resource group, storage and network) for production is available under the Agreement.

3. Data is not moved outside Japan without the Customer's prior written consent.

Article 21 (Access by Pioneerwork)

1. Pioneerwork staff access Customer data only to provide the Service, assure quality (annotations), handle incidents and give support.

2. Access is limited to authorised staff and audit-logged.

3. Data is not disclosed to third parties except as required by law, in which case the Customer is notified unless prohibited.

Article 22 (Deletion and return)

1. The Customer may request deletion of any job at any time; Pioneerwork completes deletion within 10 business days and confirms it.

2. On termination, Output Data is returned in a common format (CSV etc.) on request and all Customer data is deleted within 30 days; backup copies expire with the backup retention period (at most 35 days).

3. No data is retained after deletion except where law requires.

Article 23 (Personal data)

The Service processes product documents and is not intended for personal data. Information about Users (name, e-mail, activity) is handled under Pioneerwork's privacy policy at www.withzag.com. Where personal data is contained in documents, Pioneerwork acts as the Customer's processor under Japan's Act on the Protection of Personal Information and follows the Customer's instructions.

Article 24 (Confidentiality)

Each party keeps the other's technical and business information (including Input and Output Data, prices and unreleased product information) confidential, does not disclose it without prior written consent and uses it only for the Service. Standard exclusions apply (public, previously held, lawfully obtained from third parties, independently developed). The obligation survives for 5 years after termination; a separate NDA prevails if executed.

Article 25 (Sub-processors)

Sub-processors are listed in Annex 1. Additions or changes are notified 30 days in advance and the Customer may object on reasonable grounds. Pioneerwork imposes equivalent data-protection obligations on sub-processors and remains responsible for them.

Article 26 (Technical measures)

1. All traffic is encrypted with TLS.

2. Stored data is encrypted at rest (Azure Storage, AES-256).

3. Access is authenticated by access code or user account; administrative functions are restricted to administrators.

4. Audit log: administrator corrections, column operations, deletions and Pioneerwork annotations are recorded with actor, time and before/after content.

5. Secrets (API keys etc.) are held in Azure secret management and never written to source code, logs or screens.

Article 27 (Organisational measures)

Access to Customer data is limited to staff with a business need, bound by confidentiality. Production changes are applied after verification in a staging environment, with a record.

Article 28 (Vulnerabilities)

Pioneerwork monitors vulnerability information for the software the Service depends on and remedies critical vulnerabilities promptly.

Article 29 (Security incidents)

On becoming aware of an actual or suspected leak, loss or corruption of Customer data, Pioneerwork notifies the Customer as a rule within 72 hours with known facts, scope and status, investigates the cause and reports remediation.

Article 30 (Audit and security questionnaires)

The Customer may, within reason, request answers to security questionnaires and supporting documents. On-site audits are subject to conditions in the Agreement.

Article 31

The Service software, extraction and classification methods, screen designs, documentation, trademarks and generic attribute taxonomies (containing no Customer-specific information) belong to Pioneerwork or its licensors.

Article 32

Pioneerwork grants the Customer a non-exclusive, non-transferable right to use the Service under these Terms and the Agreement. The Customer shall not copy, modify, reverse-engineer, resell or lend the Service.

Article 33

Pioneerwork may use feedback free of charge to improve the Service; Article 24 applies to confidential information in feedback.

Article 34

Unlawful use; registering documents that infringe third-party rights; overloading, unauthorised access or vulnerability probing without prior approval; using Output Data to train competing AI models; sharing access codes; obstructing operation.

Article 35

Pioneerwork may restrict or suspend use after notice (or with notice afterwards in emergencies) for breaches of Article 34 or to protect the Service.

Article 36

Fees for production use (initial, monthly, usage-based, additional features) and payment terms are set in the Agreement. Trials are free.

Article 37

Term and renewal are set in the Agreement.

Article 38

The Customer may terminate with the notice period in the Agreement. Pioneerwork may terminate for material breach not cured within a reasonable cure period.

Article 39

On termination, Article 22 applies. Articles 19, 24, 31 and Chapters 10 and 11 survive.

Article 40

1. Pioneerwork uses commercially reasonable efforts to provide the Service according to the specification in the Agreement.

2. Targets, measurement and remedies for extraction accuracy and coverage are set in the Agreement (SLA).

3. No other warranty, express or implied, is given as to completeness, accuracy, fitness for a particular purpose or non-infringement.

Article 41

Availability targets, maintenance windows and recovery objectives are set in the SLA.

Article 42

1. Except for wilful misconduct or gross negligence, Pioneerwork's liability is limited to direct and ordinary damages, capped at the fees paid by the Customer for the Service in the 12 months preceding the event.

2. Pioneerwork is not liable for lost profits, indirect or special damages, or data-recovery costs not attributable to Pioneerwork.

3. For trial use, Pioneerwork is not liable except for wilful misconduct or gross negligence.

Article 43

Pioneerwork is not liable for delay or failure caused by events beyond its reasonable control (natural disaster, war, terrorism, changes in law, widespread platform outages, telecommunications failures).

Article 44 Neither party assigns or pledges its position, rights or obligations without the other's prior written consent. Article 45 Each party represents it is not, and is not related to, an anti-social force; breach permits termination without notice. Article 46 The Customer complies with applicable export-control laws. Article 47 Invalid provisions do not affect the rest. Article 48 These Terms and the Agreement are the entire agreement. Article 49 Governed by the laws of Japan; the Tokyo District Court has exclusive jurisdiction in the first instance. Article 50 Matters not provided for are resolved in good-faith consultation.

Contact: see www.withzag.com.