Papmil pdf workflows
Menu
How it works Pricing FAQ Login Process a PDF

Privacy

Your document's path, plainly.

Privacy language should make the system easier to understand, not make the infrastructure disappear.

The short version

Papmil is hosted. Your files do not stay on your device: they are uploaded to AWS S3 and processed by Papmil's worker. AI inference for splitting and naming runs through Ollama on that worker, not through an external AI API. File copies are automatically deleted 30 days after the workflow finishes; workflow metadata remains.

1. Information Papmil handles

Account information

Papmil stores your email address, authentication records, account status, plan, credits, timezone, and security-related sign-in information. Passwords are stored as cryptographic hashes through Devise; Papmil does not store a readable copy of your password.

Document and workflow information

To provide the service, Papmil handles uploaded PDFs, generated PDFs, filenames, file sizes, workflow and step status, timing, processing parameters, errors, and result metadata.

Technical information

The application may process IP address, browser and device information, visited pages, session data, Turnstile results, and operational logs for security, reliability, and product improvement.

2. How documents are processed

  1. Your browser uploads a PDF over HTTPS.
  2. The file is stored in AWS S3.
  3. A queued processing job tells Papmil's worker which operation to run.
  4. The worker downloads the input into a temporary job directory.
  5. Ghostscript, Tesseract, PDF libraries, and Ollama, where needed, process the file.
  6. Outputs return to AWS S3 and become result files in the Papmil application.
  7. The worker removes its temporary job directory.

3. AI processing

The current smart splitting and smart naming operations send extracted document content, and optionally page images, to an Ollama model running on Papmil's processing worker. The current pipeline does not send document content to OpenAI, Anthropic, Google Gemini, or another external AI API.

This is “local” to Papmil's processing machine, not local to your computer.

4. Service providers and infrastructure

ServiceProviderPurpose
Object storageAmazon Web Services (S3)Uploaded, intermediate, and result files
Processing queuesAmazon Web Services (SQS)Sending jobs to and from the processing worker
Transactional emailAmazon Web Services (SES)Account and workflow-completion messages
Bot protectionCloudflare TurnstileProtecting public forms from automated abuse
Application and databasePapmil-operated hostAccounts, workflow records, application, and database
Product analyticsAhoy in Papmil's databaseUnderstanding visits and product use
Website analyticsanalytics.ddventures.coProduction website statistics; exact deployment and retention to confirm

5. Retention

Source files, intermediate service objects, and result files become eligible for automatic deletion 30 days after a workflow finishes. Workflow, step, credit, and non-file result records remain so users can see processing history and Papmil can maintain aggregate proof-of-concept counts.

Account, analytics, security-log, backup, and operational-log retention periods must be confirmed before this draft becomes the production policy.

6. Cookies and sessions

Papmil uses the session mechanisms required to keep users signed in and protect account actions. The exact cookie and analytics inventory should be verified in production and documented here before publication.

7. Your choices and rights

Depending on where you live, you may have rights to access, correct, export, restrict, object to, or erase personal data. Contact [email protected] to make a privacy request. The exact account-deletion workflow and any legally required identity details must be added before publication.

8. Changes and contact

This draft will change as the product and processor inventory become clearer. Privacy questions can be sent to [email protected].