Rolebooks

Privacy policy

Document
Legal
Updated
November 4, 2026
Applies to
Website, portal, API

This policy covers the Rolebooks website, portal, drafting chat, API and connected-agent access. Rolebooks is the service brand. Contracting provider: [Legal name and postal address to be completed before customer contracting]. Contact our privacy lead at contact (at) getrolebooks.com. Read this alongside our Terms of Service. Business customers can ask for our data processing agreement.

We are accountable for the personal information we handle under applicable Canadian privacy law, including the Personal Information Protection and Electronic Documents Act (PIPEDA). For content a business puts in its workspace, that business decides the purpose and who may access it; we process it to provide the service on its instructions. Ask your employer or workspace administrator about its own practices.

Information the service collects

We collect account and business contact information (such as names, work emails, company and assigned roles), sign-in and session records, permissions, support correspondence, subscription status, billing identifiers and payment/AI-usage records. Stripe handles payment details; we do not store full payment-card numbers.

Customer content includes source documents, rolebooks, drafts, chat messages, tool requests and results, uploaded files and extracted text, selected quotes, discussions, review copies, decisions, approver identity and timestamps, change records and file hashes. An upload can contain personal information even if its file name does not. Supply only what is necessary; remove credentials, customer lists and employee/personnel details first.

Requests also involve technical information such as IP address, browser/device information, time, request status and security events. Essential cookies support signed-in sessions and security; the marketing site's visit attribution is described below. This site has no analytics or advertising trackers. We do not sell personal information or share it for targeted advertising.

Why we use information

We use information to provide and secure accounts; restrict access to the right workspace and roles; capture, draft, check, review and distribute manuals; answer support and sales inquiries; deliver sign-in and service emails; bill and reconcile usage; prevent abuse; maintain approval evidence; and meet legal obligations. We do not use one client's private material to draft another client's manuals or publish it as marketing without permission.

You or your business provide the authority and any notices or consents needed for staff and other people mentioned in customer content. You can decline optional processing or withdraw consent by contacting us; we explain the effect on the service and any legal or record-retention exception. Using the service is not blanket consent to optional meaning search.

Uploads and AI-assisted drafting

Draft attachments are private to the client's workspace, never published as attachments, and used only to draft that client's manuals. They can be added in the drafting chat or by an authorized agent on your computer through MCP. Supported inputs are CSV/TSV, JSON, plain text, Markdown, PDF, DOCX, PNG and JPEG, subject to service limits and content checks. We extract text where supported; accepted images have camera metadata stripped. This does not remove personal information visible in an image or document.

AI drafting and triage may send your instructions, relevant discussion or manual context, extracted attachment text and tool results to the hosted model route. Hosted chat uses OpenRouter and the model endpoint it selects. Our server sets zdr: true, data_collection: "deny" and allow_fallbacks: false. OpenRouter interprets ZDR routing together with account/guardrail settings, which may permit Safety Retention endpoints; the request flags alone do not establish strict zero retention. Provider safety screening, flagged-content review and legal preservation exceptions may also apply. We make no blanket no-retention promise. Confirm current endpoint and account settings with us before sending restricted material; Rolebooks separately keeps service records as described below.

AI can make mistakes. Proposed wording passes checked edits, review copies and customer approval before issue; AI does not itself authorize publication. HR/personnel and other agreed AI-excluded documents must not be sent for AI drafting or triage; excluded work is drafted manually. Ask us to record exclusions before sharing material. Do not put sensitive information in chat expecting an exclusion to redact it afterward.

If you connect Claude Desktop, Claude Code or another external agent, you choose that application and its model provider. Information the agent reads or sends may also be processed under that provider's separate terms and settings, outside our hosted OpenRouter routing controls. Grant only necessary access and revoke the connection when no longer needed.

Find by meaning retrieves approved manual sections; it does not generate answers. It stays off for real clients until a separate, explicit client consent and settings/retrieval checks are completed. Signing the service terms alone does not enable it. Ordinary reading and keyword search remain available.

When enabled, Cloudflare AI Search receives approved, non-withdrawn role-specific section text, coded section/role/edition identifiers and readers' search words, and stores a derived index including chunks and Workers AI embeddings. Search requests do not include the reader's name or email. Excluded source modules and ancestors containing them stay out of AI Search but remain available to authorized readers and keyword search. Readers should not type personal details in queries.

The separate consent records current provider settings, logging/retention checks, exclusions and staff notice. Withdrawal disables retrieval and starts background deletion of derivatives; edition withdrawal deletes that edition's derived items. Offboarding includes disabling search and purging the index. Deletion is checked to completion; we do not promise a fixed provider deletion deadline. Retained source editions are separate. We make no Canadian-only processing or blanket no-retention promise.

Service providers and other disclosures

We use these providers for the stated functions, with access limited to what those functions require:

  • Cloudflare: website/portal/API hosting, Workers, D1 databases, Durable Objects for drafting workspaces, private R2 file storage, security (including Access and Turnstile), Email Service for outbound email, and optional AI Search with Workers AI embeddings. Cloudflare privacy policy; customer DPA.
  • OpenRouter and its eligible model-serving provider: hosted drafting/triage inference under the privacy route above. The model-serving endpoint can vary within those restrictions; request current model/endpoint details before sending restricted material. OpenRouter privacy policy; ZDR information.
  • Stripe: subscription checkout, billing, payment processing and related fraud/compliance records, when billing is used. Stripe privacy policy.
  • GitHub: private client library repositories, source/version history, builds, review/release automation and release records. GitHub privacy statement.

Cloudflare delivers service email to the recipient's email system. Correspondence is also held in our business inbox. Your own email, agent and other systems are governed by their providers' terms. We may disclose information when required by law, to protect rights or security, or in a business transfer subject to continued confidentiality and notice where required. We do not authorize providers to use customer content for unrelated advertising or model training.

Processing outside Canada

We and these providers may process or store information in Canada, the United States and other countries. It may be accessible to courts, law enforcement or national-security authorities under those countries' laws. We use contractual safeguards and restricted access, but do not promise Canadian residency. Contact our privacy lead for information about foreign processing and provider safeguards.

Retention, cancellation and deletion

Draft attachments: kept while the account is active and deleted within 30 days after cancellation or on written request, except where cited in approved change provenance. Those cited attachments are kept with the edition record. A file cited in recorded provenance cannot be removed through the attachment-delete tool; contact us so we can determine which records must be kept. Deleting a draft attachment does not erase already-approved text or retained approval evidence.

Edition and approval history: issued editions have an indefinite storage lock. We retain editions, all review copies (including superseded or undecided copies), change records, approvals, discussions and the client repository indefinitely to preserve what was reviewed, approved and issued. Cancellation or a deletion request does not automatically delete these records. Corrections to issued content are made in a new edition, not by rewriting history. Cited approved-change provenance is retained with that history. This is a specific exception to deletion, not a claim that all uploads are kept forever.

Access and export: after a subscription ends or becomes delinquent after payment retries, access is read-only for 30 days, then portal access and notifications stop. Request an export in that window; we gather it manually. Removing access is not deletion. See cancellation and export terms.

Other records: account, correspondence, security, payment and usage records are kept as needed for the service, financial/legal obligations and disputes. Hosted inference replay chunks are removed by hourly housekeeping after a completed step is more than an hour old; step identifiers, digests, provider generation identifiers and cost/audit records remain. That short replay period is not a deletion deadline for chat, drafts or attachments. Provider backup, security and billing records follow applicable terms; we explain applicable retention exceptions when responding to a request.

Security and confidentiality

We use encrypted transport, authenticated workspace/role access, restricted maintainer access, private file storage and checked approval/release workflows. Access is limited to authorized personnel and providers who need it for the service. No system is completely secure. Protect sign-in email, devices and agent/API credentials; tell us promptly about suspected unauthorized access. We assess security incidents, keep required records, and notify affected businesses, individuals and regulators as applicable law requires.

What the contact form collects

The details you type in: your name, email address, company name, type of business, number of employees, and anything you add about what you’d like to discuss.

It also sends how you found us: the campaign tags (UTM tags) in the link you followed, the website that sent you, and which option you chose if you came from a link such as “Tell us about it”. Your browser keeps these for the current visit only, so they can go with your message.

What we do with it

Your submission is emailed to our inbox and kept there so we can reply and prepare for the demo. It is not saved to a website database, and you are not added to a mailing list.

What the org chart builder collects

To draw your chart: your company name, brand colour, paper size, role titles, who each role reports to, any departments you add, and your logo if you upload one. To email it: your name, email address, phone number, type of business, and number of employees, plus the same visit details as the contact form.

What we do with your chart

Your chart details are used to draw the preview and the PDF while you wait. Your logo and the PDF are not stored: once the chart is drawn and emailed, we keep no copy of either.

When you ask for the PDF, we email it to the address you give, and email your contact details and your chart’s roles and reporting lines (not the logo) to our inbox. We keep that so we can follow up with you by email or phone about your chart and our service. Tell us any time if you’d rather we didn’t. You are not added to a mailing list.

So the builder can’t be used to send repeated emails to someone, we keep a scrambled, one-way code made from your email address (not the address itself) for one day to count how many charts it has been sent.

Spam protection and hosting

Both forms use Cloudflare Turnstile, a security check that processes information from your browser and your IP address to block automated spam. Your IP address is also used to limit repeated submissions. Cloudflare hosts this site, processes the requests your browser makes to it, and delivers our emails. See Cloudflare’s Turnstile privacy addendum and privacy policy.

This site has no analytics or advertising trackers.

Access, correction, deletion and complaints

Write to the Rolebooks privacy lead at contact (at) getrolebooks.com to ask about processing or foreign providers; request access to or correction of personal information; request deletion or withdraw consent; or make a privacy complaint. You can also use our contact page. We verify identity and authority proportionately and respond to access requests within 30 days, subject to lawful extensions and exceptions that we explain. We do not need your password, sign-in code or complete card details.

If we process information for your employer, we may coordinate the request with its authorized administrator. We explain any refusal or record-retention exception, including immutable edition history, and available correction or restriction options. You may challenge our response with our privacy lead and, if unresolved, contact the Office of the Privacy Commissioner of Canada. Applicable statutory rights are not waived by these policies.

Policy changes

We update the date above when this policy changes and notify account contacts of material service changes before they take effect where practicable. Where a new purpose needs consent, we seek it before that processing begins. Separate Find by meaning consent is renewed for material changes covered by that record.