Your manuscripts stay private.
Your draft is the most personal thing you own. We built PlotLens so that trusting us with it is a decision you can check, not one you have to take on faith.
Reviewing PlotLens for your organization? Hosting, storage, backup, retention, monitoring and AI-provider statements last checked against production on 2 November 2026. Anything not listed as checked was not re-verified on that date.
PlotLens does not train models on your work. Yours to export. Yours to delete.
The three things that matter most
Your work is never training data
PlotLens never uses your manuscripts, characters or worlds to train or fine-tune AI models. Provider terms are summarized in the AI processing section below, where we have them. Your book stays your book.
Encrypted, with access checked every time
Encrypted with TLS on the way to us and AES-256 once it lands. Access to your work is checked against your account on every request, and the database adds row-level protection on part of its schema. The controls reference says which part.
Leaving is as easy as arriving
Export your content using the available export tools. Delete a document, a project or your whole account from settings and it leaves the product. Backup copies expire on their own schedule. Two manual database snapshots from the move to AWS remain scheduled for deletion on 1 December 2026; the retention section has the details.
For security reviewers
Controls reference
Everything below describes what we found in place when we checked, not what is planned. Where a control does not exist yet, or we cannot yet document it, it says so. Anything genuinely planned is in the roadmap at the foot of this page.
Sections
Tenant isolation and access management
| Control | How it works |
|---|---|
| Tenant data isolation | PostgreSQL row-level security (RLS), bound to a single tenant per database session, is enforced on a subset of tables. Other tables rely on application authorization or contain global operational data. A read-only catalog measurement on 7 November 2026 found 103 public-schema tables: 60 had row-level security enabled and forced, and 43 did not have it enabled. Of the 43, 42 match the repository’s declared application-scoped or global/operational exemptions; the other is the schema-version table. This is a catalog census, not a complete access-control assessment. PlotLens has not yet published a complete, independently verified account of RLS coverage; ask the security team for the current inventory. |
| Background processing | Background workers are configured to connect with a separate, scoped database role rather than the application role, and the user-facing services are not given the worker credentials. |
| Authentication | Delegated to Clerk. PlotLens never stores account passwords. The API gateway authenticates protected requests before forwarding them to application code, using signed Clerk sessions or supported API keys. Public endpoints have explicit authentication exemptions. |
| Authorization | Four team roles (owner, admin, editor and viewer) are evaluated server-side on every request, never trusted from the client. |
| Administrative access | Platform-administrator access is a persisted role. It is never derived from a client-supplied email address or header. A documented access review of production systems is not yet available; ask the security team for the current position. |
| Audit logging | Security-relevant events are recorded with actor, action, resource, event category, source IP address and timestamp where available. On 7 November 2026, the production API database role had neither UPDATE nor DELETE permission on the audit table, and update/delete guards were enabled. This does not establish immutability against privileged database operators. Some event writes are best-effort, so this is not a guarantee that every operation produces a record. |
Encryption
| Control | Detail |
|---|---|
| In transit | The application endpoint negotiates TLS 1.2 and TLS 1.3. Connections from the application to the database require TLS. Traffic between services inside the cluster is not separately encrypted; it stays on the cluster’s private network under default-deny network policies. |
| At rest | On AWS, in the US East (Ohio) Region: the PostgreSQL database is encrypted with a customer-managed key in AWS KMS, manuscript object storage uses S3 server-side encryption (AES-256), and the disks behind the vector database and cache and the shared model volume are encrypted. |
| Per-project envelope encryption | Manuscript bodies uploaded from August 2026 onward are encrypted by PlotLens itself, in addition to the storage layer above, before the bytes reach storage. Each project gets its own key-encryption key in Azure Key Vault. The body is encrypted with AES-256-GCM under a data key belonging to that project, and the data key is stored only in wrapped form. Unwrapping it is a call to the vault, and the key-encryption key’s private material never leaves the vault. AWS workloads reach the vault over the network. |
| What that covers, and what it does not | Files written before August 2026 are covered by the storage-layer encryption above and nothing further. The cutover is forward-only: no re-encryption pass runs over older manuscripts. Envelope encryption protects stored file bodies. It does not cover text-derived data in the database or the search index, and it does not by itself delete anything. |
| Key management | PlotLens does not claim FIPS 140-3 validation or hardware-security-module backing for its key store. Customer-managed keys (your vault, your key) are not offered today. Ask the security team about key-handling details. |
| Secrets | Application secrets and provider credentials are held in AWS Secrets Manager and synchronized into the cluster by the External Secrets Operator. CI scans the repository for committed secrets and fails the build when it finds one. |
Two layers, stated separately on purpose: storage-layer encryption on AWS, and per-project envelope encryption for manuscript bodies written from the cutover onward. Neither layer is independently attested. PlotLens holds no third-party certification or audit report, and nothing on this page should be read as external compliance attestation or FIPS coverage of these controls. If customer-managed keys are a requirement for your review, raise it before purchase.
Data residency
| Scope | Location |
|---|---|
| Application and primary data | AWS us-east-2 (Ohio, United States). Application compute (Kubernetes on Amazon EKS), the primary PostgreSQL database (Amazon RDS), manuscript object storage (Amazon S3) and the vector database and cache (running in the cluster) are in this Region. Production moved here from Azure on 28 October 2026. |
| Backups | Database backups are kept in the same Region (us-east-2). We found no cross-Region copy. |
| Encryption keys | Per-project key-encryption keys are held in Azure Key Vault, Azure centralus (Iowa, United States). |
| AI chat inference | Amazon Bedrock (Amazon Web Services), US (us-east-2). Requests use a US cross-Region inference profile configured from us-east-2, so they may be processed in US AWS Regions. PlotLens attaches no Bedrock Guardrails. For the OpenAI models PlotLens uses on Bedrock (other than the gpt-oss models AWS lists as exempt), AWS’s Bedrock abuse-detection documentation and Service Terms §50.12.2 state that classifier-flagged inputs and outputs may be retained for up to 30 days for abuse detection and that AWS may review them. PlotLens has not confirmed zero data retention (ZDR) for its AWS account. Sources checked 2026-11-03: abuse detection; AWS Service Terms §50.12.2. |
| AI embeddings | The Azure OpenAI resource is provisioned in centralus, but its model deployments use Azure’s Global Standard deployment type. Microsoft may therefore process an individual inference request in a data center outside the United States. Data the service stores at rest stays within the resource’s geography. |
| Marketing site | This website is hosted separately from the product, on AWS (us-east-1, served through a content delivery network). It holds no customer content. |
| Company | PlotLens is operated from the United States by Family Friendly, Inc. (Delaware, US). |
If US-only inference is a hard requirement for your review, raise it before purchase. The routing above is the honest position today, not a US-pinned guarantee.
Retention and deletion
What happens to data after you delete it, including retained database snapshots from the move from Azure to AWS. Removing data from the live product does not immediately remove retained copies. The table distinguishes scheduled expiry, planned removal and the retired Azure file storage.
| Data | Retention | Mechanism |
|---|---|---|
| Documents and manuscripts | Until you delete them | Deleting a document removes it from the product immediately. In object storage the delete leaves earlier versions of the file in place for 30 days, after which they expire. Copies in backups, snapshots and logs follow their own windows. |
| Account, projects and story bible | Life of the account | Deleting your account removes your account and content from the live product. We do not state a single deletion deadline, because copies in the backups, logs and migration copies below persist until they expire or are removed. |
| Project encryption keys (crypto-shredding) | 30-day grace period | Deleting your account also destroys the key-encryption key of every project you solely own. The key stops being usable immediately, and is permanently purged after a 30-day grace period. Reversal inside that window is not self-service and is not guaranteed. By design, once the key is purged the encrypted file bodies written from August 2026 onward can no longer be decrypted, and their copies in backups hold only the wrapped data key. This design has not yet been verified by a deletion drill in production, so do not treat it as a verified guarantee. It does not reach files written before August 2026, or text-derived data in the database and search index, which are deleted by other means. Derived data in backups persists until they expire. Projects owned by a team are deliberately not shredded when one member leaves. |
| Database backups | 30 days | Automated database backups are kept for 30 days and roll off automatically. Backups cannot be selectively edited, so a deleted record persists inside them until the window passes. |
| Migration snapshots of the database | Deletion planned 1 December 2026 | Two manual database snapshots were taken during the move to AWS, one before the cutover on 28 October 2026 and one on 31 October 2026 when the Azure database was retired. Manual snapshots do not expire on their own. Both are encrypted, and both contain data as it stood on those dates, including data you may have deleted since. PlotLens plans to delete both on 1 December 2026. |
| Retired pre-migration Azure file storage | Account removed 2 November 2026 | The Azure storage account that held the pre-migration copy of manuscripts, entity images, exports and vector-index snapshots was removed on 2 November 2026. Read-only checks of the production subscription on 7 November 2026 found no active account or recoverable deleted-account entry for it. The retained database snapshots above are separate copies and remain scheduled for removal on 1 December 2026. |
| Current application logs | 30 days | Application logs are kept in AWS CloudWatch Logs for 30 days. They are not designed to hold manuscript content, but error messages and stack traces are not screened for it, and they can contain identifiers such as user, project and document IDs. Destroying a project's encryption key does not affect them. |
| Object-storage previous versions | 30 days (manuscripts), 1 day (exports), one week (vector-index snapshots) | The S3 bucket is versioned. S3 has no separate soft-delete feature: deleting or replacing a file keeps the previous version, and a lifecycle rule expires it after the stated window. AWS applies lifecycle rules on its own schedule, so expiry can follow the window by up to a day or so. |
| Audit logs | Beyond account deletion | Retained where required for security and legal obligations. The production API database role cannot update or delete audit records. Privileged database operators remain outside that guarantee. The retention period is not published. Audit records include actor and action metadata. Some existing event details can include Story Bible rule names, changed values and entity snapshots used for Undo. These records are not covered by manuscript-file envelope encryption or key destruction. |
| Analytics events | Per sub-processor policy | Product analytics follow your consent choice. A server-side embedding timing event is sent independently of user analytics consent; the sub-processor section lists its fields. |
The product does not offer self-service recovery of deleted content. Retained database backups are whole-system copies; object storage separately retains individual file versions that may remain recoverable by authorized operators until expiry. The retained migration database snapshots follow the separate removal plan above. Removing product access does not immediately erase these retained copies.
AI processing
| Question | Answer |
|---|---|
| Who processes the text? | Chat analysis (extraction and Manuscript Check): Amazon Bedrock, operated by Amazon Web Services under PlotLens’s own AWS account. Embeddings: Azure OpenAI Service, operated by Microsoft under PlotLens’s own Azure subscription. |
| Is my writing used for training? | PlotLens never uses your manuscripts to train or fine-tune any model. Microsoft’s documentation for Azure OpenAI states that prompts, completions and embeddings “are NOT used to train any generative AI foundation models without your permission or instruction” and are not shared with OpenAI (source). The Amazon Bedrock FAQ states that “AWS and the third-party model providers will not use any inputs to or outputs from Amazon Bedrock to train Amazon Nova, Amazon Titan, or any third-party models” (source). Both checked 2026-11-03. |
| Does the provider retain anything? | For embeddings, Microsoft’s Azure OpenAI / Foundry documentation (checked 2026-11-03) states that abuse monitoring may sample prompts and completions for review. Review is automated by default (including by AI models); human review by authorized Microsoft employees happens only when needed. Prompts and completions stored for human review are not used to train models. Microsoft’s current data-privacy and abuse-monitoring pages do not publish a retention period. PlotLens has not applied for modified abuse monitoring (Limited Access) that turns off human-review storage. Sources: Foundry data privacy; abuse monitoring. For analysis (chat) on Amazon Bedrock, AWS Service Terms §50.12.2 state that for certain models Amazon Bedrock stores service inputs and outputs for up to 30 days solely to detect policy violations, and may review them. AWS’s abuse-detection page states that for all OpenAI models other than the listed gpt-oss models, classifier-flagged traffic will be retained for up to 30 days. PlotLens’s configured chat models are OpenAI models on Bedrock; PlotLens has not confirmed ZDR. Checked 2026-11-03. |
| What exactly is sent? | The passage under extraction or Manuscript Check, plus the relevant Story Bible context, is sent to the chat provider. Text is sent to the embeddings provider to build search vectors. PlotLens has not yet published a field-by-field inventory of every request. |
| What about embeddings? | Text is sent to Azure OpenAI to be converted to vector embeddings for semantic search. The resulting vectors are stored in PlotLens’s own vector database, which runs in the AWS cluster in us-east-2 on encrypted disks. |
| Do PlotLens staff read my work? | A documented review of staff access to production data is not yet available. Ask the security team for the current position. |
Sub-processors
Infrastructure and AI rows checked against production on 2 November 2026. The rows for third-party business tools (authentication, payments, email, CRM, support, analytics) were not re-verified on that date. These are the third parties that may process customer data on PlotLens’s behalf.
| Sub-processor | Purpose | Data received | Region |
|---|---|---|---|
| Amazon Web Services | Production hosting: Kubernetes (EKS), PostgreSQL (RDS), object storage (S3), application logs (CloudWatch Logs), encryption keys for storage (KMS), secrets (Secrets Manager); also marketing-site hosting | All customer content and account data | US (us-east-2); marketing site us-east-1 |
| Microsoft Azure | Per-project encryption-key vault (Key Vault), identity for access to that vault, pre-migration logs, and static add-in and release-site infrastructure | Project encryption keys and key-operation metadata; application logs written before the move to AWS (the Azure log workspace has received no production logs since the move to AWS in late October 2026; its retention period is not published). The retired Azure file-storage account was removed on 2 November 2026 | US (centralus) |
| Amazon Bedrock (Amazon Web Services) | Story element extraction and Manuscript Check | Manuscript text submitted for processing | US (us-east-2); US cross-Region inference |
| Azure OpenAI (Microsoft) | Embeddings for semantic search | Manuscript text submitted for embedding | US (centralus) resource; Global Standard inference may process outside the US |
| Clerk | Authentication and identity | Email, name, authentication metadata | US |
| Stripe | Payments | Billing contact, payment method (held by Stripe) | US |
| Microsoft commercial marketplace | Subscription billing for Marketplace purchases | Subscription and entitlement metadata | US |
| Loops | Lifecycle and transactional email | Email address, product usage events; email address from the blog newsletter signup | US |
| HubSpot | Customer-relationship management (CRM) | Name, email, sign-up date, plan and trial status; campaign tags only with analytics consent. From the Contact Sales form: name, email, company, role, seats, message, whether you asked for the data processing agreement, the page you sent it from, and the HubSpot tracking cookie (only with analytics consent). No payment details or manuscript content | United States (West) |
| GitHub (Microsoft) | Support request tracking | Support request text and contact email you submit | US |
| Slack | Support request notifications | Support request text you submit; no email address | US |
| PostHog | Product analytics and operational timing | Consent-gated product usage events; a server-side embedding timing event is sent independently of user analytics consent with a system identity, model deployment, outcome, item/token counts and timing metadata | US |
| Google Analytics / Tag Manager | Marketing-site analytics (consent-gated) | Marketing-site page views only | US |
| Microsoft Clarity | Marketing-site heatmaps and session replay (consent-gated) | Marketing-site interactions only | US |
Browser analytics from PostHog, Google Analytics and Microsoft Clarity load only after you accept analytics cookies. The server-side embedding timing event listed above is sent independently of user analytics consent. It does not include manuscript text or user, project or document identifiers. None of them is designed to receive manuscript content. PlotLens does not currently run Azure Application Insights error monitoring; that resource was deleted on 31 October 2026.
Application security and SDLC
| Control | How it works |
|---|---|
| Dependency and secret scanning | Trivy scans the repository on every change. A fixable HIGH or CRITICAL vulnerability fails the build unless it is recorded as an exception in the repository, and a committed secret fails the build. |
| Container image scanning | Images are scanned with Trivy during the build. A fixable CRITICAL finding fails the build; HIGH findings are reported but do not fail it. |
| Infrastructure-as-code scanning | Trivy’s configuration scanner checks Terraform and Helm definitions on CI runs other than pull-request and merge-queue runs. Findings are informational and do not fail the build. |
| Static analysis | SonarQube analyzes pull requests. |
| Change control | Changes reach the main branch through pull requests that must pass a required CI check and a merge queue. Application releases are deployed by a CI workflow that needs an approval to run; infrastructure changes are applied by an operator. |
| Environment separation | Production runs in its own AWS account, separate from the AWS development account. |
Availability and resilience
| Control | Detail |
|---|---|
| Uptime commitment | PlotLens does not offer a contractual uptime SLA on its self-serve plans. Uptime commitments are available only under a negotiated Enterprise agreement. |
| Database durability | Amazon RDS for PostgreSQL 16, encrypted, with automated backups retained for 30 days in the same Region. A read-only check on 7 November 2026 found a single-Availability-Zone deployment; it does not provide Multi-AZ database failover. |
| Object durability | Manuscript files are stored in a versioned Amazon S3 bucket. A replaced or deleted file keeps its previous version for 30 days, then it expires. |
| Monitoring | External uptime checks cover the public website, the web app and the API endpoints. Ask the security team for monitoring and alerting details. |
| Release safety | The production deploy workflow upgrades with automatic rollback on a failed rollout and then checks the health endpoint through the load balancer. A build identifier is exposed on the health endpoint so the running version can be verified independently. |
| Status page | Not yet published. See the roadmap below. |
Incident response
| Control | Detail |
|---|---|
| Detection | External checks watch the public endpoints. Ask the security team about on-call and escalation. |
| Breach notification | If PlotLens confirms a personal-data breach affecting your data, we will notify affected customers without undue delay. Enterprise agreements can specify a fixed notification window. |
| Post-incident | Recurring root causes are written up in internal runbooks so a repeat failure can be diagnosed against a known cause. |
| Reporting an incident | Email security@plotlens.ai. |
Compliance and legal
| Topic | Position |
|---|---|
| Our role | For your manuscripts and account data, you are the controller and PlotLens is the processor. For this marketing site’s own analytics, PlotLens is the controller. |
| GDPR | Access, correction, deletion, export and objection rights are supported from account settings or by writing to privacy@plotlens.ai. |
| CCPA / CPRA | PlotLens does not sell personal information and does not share it for cross-context behavioral advertising. |
| International transfers | Primary hosting is in the United States; AI processing follows the routing and geography limits in the Data residency section above. Transfers from the EEA, UK and Switzerland rely on the EU Standard Contractual Clauses, supplied with the data processing agreement. HubSpot participates in the EU-U.S. Data Privacy Framework (with the UK Extension and Swiss-U.S. framework) and its DPA incorporates the SCCs. |
| Data processing agreement | Available on request from privacy@plotlens.ai. |
| Third-party attestations | PlotLens does not currently hold an independent third-party security certification, and has not commissioned an external security audit. This page exists so you can assess the controls that are actually in place, rather than infer them from a badge. |
| Age | PlotLens is not intended for users under 16 years of age. |
Vulnerability disclosure
| Topic | Detail |
|---|---|
| Where to report | Email security@plotlens.ai. A machine-readable contact is published at /.well-known/security.txt. |
| Acknowledgement | We aim to acknowledge a report within three business days. |
| Safe harbour | We will not pursue legal action for good-faith research that respects user privacy, avoids degrading the service, and does not access or modify data belonging to others. |
| Bug bounty | PlotLens does not currently run a paid bug-bounty program. |
Enterprise
Requirements that fall outside the self-serve plans are handled as a private offer rather than a plan tier.
Custom infrastructure and terms
Where custom infrastructure or custom terms are required, PlotLens Enterprise is available as a private offer set by contract. Contact support@plotlens.ai.
Negotiated uptime commitments
Uptime commitments, service credits and support-response targets are agreed in an Enterprise contract. They are not offered on the self-serve plans.
Data processing agreement and SCCs
A data processing agreement incorporating the EU Standard Contractual Clauses is available on request from privacy@plotlens.ai.
Security questionnaires
Send your questionnaire to security@plotlens.ai. Most of it is answerable from the controls reference above; we will complete the rest and tell you plainly where the answer is “not yet”.
On the roadmap
Listed here precisely because it is not shipped. Nothing in this table is available today, and none of it is a contractual commitment.
| Item | Status | Target |
|---|---|---|
| SSO and SAML | Not implemented today. | Not yet scheduled |
| Public status page | Planned. Availability is currently communicated by email. | Not yet scheduled |
Security questions?
Ask us anything about how we handle your work. If we cannot substantiate an answer, we will say so.