Terms and Privacy

Last updated: June 4, 2026

This document covers both the Terms of Service and the Privacy Policy for Pragmus (pragmus.ai). It explains what the service is, what we commit to, and how we handle your data. We've written it in plain language because we think you should actually be able to read it.

Part 1 — Terms of Service

1. Who we are

Pragmus is operated by pragmus.ai. You can reach us at [email protected] for any questions.

2. Private beta

Pragmus is currently in private beta. During this period:

• The service is provided on a best-effort basis and may change significantly
• Features, data structures, and functionality may be modified, discontinued, or replaced without notice
• Data retention is not guaranteed during the beta period. We will make reasonable efforts to notify you and provide access to your data before any changes that would affect its availability
• In the event of significant service changes, shutdown, or migration, we will make best efforts to give you adequate notice and a means to export your data

By joining the waitlist or using the service during beta, you acknowledge these conditions.

Joining the waitlist does not guarantee access to the service. Pragmus reserves the right to approve or decline waitlist applications at its sole discretion. We will make reasonable efforts to notify applicants of their status.

Pragmus reserves the right to suspend or terminate any beta account at its sole discretion, at any time, with or without notice. Reasons for termination may include but are not limited to:

• Violation of these terms — including submitting unlawful, infringing, or prohibited content
• Abuse of the service — including excessive usage that disrupts the platform or other users
• Providing false or misleading information during signup
• Attempting to reverse engineer, scrape, or exploit the service
• Behavior that poses a security or legal risk to Pragmus or other users
• The beta program reaching capacity or being discontinued

Upon termination, we will make reasonable efforts to provide you with access to your data for a period of 30 days.

3. What the service is

Pragmus is a knowledge interpretation layer. It ingests content you submit — decisions, notes, agent output, session transcripts — and produces a structured knowledge graph on top of that content.

The service is designed to interpret information, not to reproduce it verbatim. Submitted content is processed by AI to extract, classify, and organize meaning into navigable knowledge items, topics, and narratives. The output is a best-effort interpretation of the submitted content — it is not a faithful reproduction, a system of record, or an authoritative source.

Important: the knowledge graph generated by Pragmus may be incomplete, reworded, or otherwise different from the original submitted content. Do not rely on Pragmus as a system of record or as a source of truth for original content. Verify extracted knowledge before relying on it for decisions or agent instructions.

If verbatim storage is required, use the document attachment feature. Documents are stored as-is and are not processed through the interpretation pipeline. Pragmus is not a document storage system — the document feature exists to allow users to attach source references to interpreted knowledge items, not as a primary storage mechanism.

4. What the service does automatically

The following actions are performed automatically by Pragmus without explicit user confirmation per action:

• Extraction — submitted content is processed by AI and decomposed into structured items
• Classification — items are classified by type, subject, and theme
• Organization — items are routed into navigable containers, and topical groupings emerge from automatic clustering
• Summarization — narratives and summaries are generated on top of accumulated items to provide navigable structure
• Relationship detection — connections between items are detected and surfaced automatically
• Resolution proposals — when newer content closes, contradicts, or invalidates older content, the system surfaces a proposal for the user to confirm or dismiss. Pragmus does not apply resolutions automatically.

All automatically generated content — groupings, summaries, relationships, and resolution proposals — is reversible by the user. No permanent action is taken without explicit user consent.

5. What users control

Users retain full control over:

• Resolution review — proposed resolutions (closed, invalidated, contradicted) can be confirmed, dismissed, or reversed at any time
• Editing, restoration, and deletion — users can edit, archive, restore, delete, or reorganize content within projects they belong to, subject to the permissions of their role. Archived or trashed content can be recovered; resolved items can be reopened.
• Milestones — users define era boundaries; pre-boundary items remain searchable as snapshots
• Sandboxes — users isolate parallel graphs inside one project; items in a sandbox cluster only with each other
• Read-only share links — project owners may opt into generating read-only links that grant access to project content without an account. These links act as bearer tokens: anyone holding the URL can view the content. The tokens are unguessable but provide no further gating. By opting in, the user accepts responsibility for how and where the URL is distributed and for any consequences of its disclosure. Pragmus is not responsible for content accessed through a share link the user has chosen to generate. Links can be revoked at any time.
• Project deletion — project owners can permanently delete a project and all its contents, including all submitted content, extracted items, generated summaries and structure, attached documents, and source records associated with that project
• Account deletion — users can delete their account at any time

Permanent deletions are irreversible. Pragmus does not take permanent action on user content without explicit user instruction.

6. MCP usage and agent-submitted content

Pragmus is accessible via the Model Context Protocol (MCP), which allows AI agents to read from and write to your knowledge base on your behalf. When you connect an agent to Pragmus and grant it write access, that agent may submit content autonomously — including decisions, observations, outputs, or any other data it determines is relevant.

Pragmus does not moderate, filter, or validate the content of MCP submissions. Submissions are processed by the same automatic interpretation pipeline as any other ingest (see §4). The user is solely responsible for:

• Configuring what agents are authorized to submit
• Ensuring that agent-submitted content does not include sensitive, confidential, or unintended data
• The actions and outputs of any agent they connect to Pragmus

Project owners may also issue programmatic credentials (e.g. API keys) that grant non-interactive access to a project. Owners are responsible for distributing these credentials and for the actions taken by any agent or system using them.

If an agent submits data that you did not intend to share — including personal data, third-party information, or proprietary content — Pragmus bears no responsibility for that submission. You retain the ability to delete any content from your knowledge base at any time.

7. Documents and attachments

Pragmus allows users to attach documents to knowledge items as reference evidence. Attached documents are stored as-is and are not processed through the extraction or classification pipeline. Pragmus does not read, analyze, or extract content from attached documents.

You are responsible for:

• Ensuring you have the right to store and share any documents you attach
• The accuracy, legality, and appropriateness of attached content

Pragmus makes no representation about the content of attached documents. Documents are accessible only to contributors with appropriate permissions on the project they belong to, or via a read-only share link the owner has chosen to generate.

8. Disclaimer of warranties

The service is provided "as is" and "as available" without warranties of any kind, express or implied.

Pragmus does not warrant:

• The accuracy, completeness, or fidelity of knowledge items extracted from ingested content
• That the knowledge graph accurately or fully represents the original submitted content
• That automatically generated groupings, summaries, or relationships are correct or meaningful
• That the service will be uninterrupted, error-free, or available at any given time
• That the service is fit for any particular purpose

The interpretation pipeline is AI-driven. Extracted items may be incomplete, reworded, misclassified, or otherwise different from the original submitted content. Users are solely responsible for verifying that extracted knowledge accurately reflects their intent before relying on it for decisions or agent instructions.

9. Limitation of liability

To the maximum extent permitted by applicable law, Pragmus shall not be liable for any indirect, incidental, special, consequential, or punitive damages arising from your use of the service, including but not limited to:

• Decisions made based on inaccurate, incomplete, or misrepresented knowledge items
• Actions taken by AI agents acting on content stored in Pragmus
• Loss of data during the beta period
• Service interruptions or changes

During the private beta period, the service is provided free of charge. Pragmus's total liability to you for any claim arising from use of the service shall not exceed the amount paid by you in the twelve months preceding the claim. During beta, where no fees are charged, this amount is zero.

10. User responsibilities

By using Pragmus you agree to:

• Use the service only for lawful purposes
• Not submit content that infringes third-party rights, including intellectual property or privacy rights
• Not use the service to store or process content you do not have the right to use
• Take responsibility for what your agents submit on your behalf via MCP
• Verify extracted knowledge before relying on it for consequential decisions

11. Governing law

These terms are governed by applicable law. Any disputes arising from use of the service shall be resolved through good faith negotiation between the parties before pursuing any other legal remedies.

Part 2 — Privacy Policy

12. What we collect

Waitlist and account data When you join the waitlist or create an account, we collect your name and email address. This is handled through our authentication provider.

Project content Once you have access to Pragmus, any content you submit — decisions, notes, agent output, session transcripts — is processed and stored as structured knowledge. This content belongs to you.

We do not collect payment information directly. Payments are handled by a third-party payment processor.

13. How we use your information

• Waitlist: to notify you when access is available
• Product: to provide the knowledge structuring service — extracting, classifying, and organizing content you submit
• Communication: to send product updates and important notices related to your account

We do not sell your data. We do not use your data for advertising.

14. How content is processed

When you submit content to Pragmus, it is sent to third-party AI providers for real-time extraction, classification, and semantic organization. These providers process content solely to deliver the service.

We use AI providers that contractually prohibit using customer inputs for model training. Content is processed in real time and is not retained by these providers beyond what is necessary to complete each request.

The text sent to AI providers consists of the item headers, bodies, and footers you submit, along with existing item text that may be re-sent when newer submissions need to be compared against it for relationship and resolution proposal detection. Item text is also sent to a separate embedding model provider for semantic indexing.

Attached documents are not sent to AI providers — they are stored as-is and bypass the interpretation pipeline. Read operations — browsing items, topics, areas, or searching the knowledge base — also do not involve AI providers; they are served directly from our database.

Structured output — the knowledge items, topics, and summaries — and any attached documents are stored on our infrastructure in the European Union.

15. Analytics

Site analytics: Pragmus uses Cloudflare for DNS, content delivery, and basic traffic analytics. Cloudflare processes request metadata — including IP addresses, page URLs, and browser type — at the network level to deliver the service and provide aggregate traffic data. Cloudflare Web Analytics is cookieless and does not track individual users across sessions.

Product analytics: we track feature usage and interactions within the app to understand how the product is used and to improve it. Events are sent server-side and may include a pseudonymous user identifier alongside the action taken (for example, which MCP tools were called or which features were used). We use an EU-based product analytics provider for this purpose. We do not use advertising analytics or session recording tools.

16. Subprocessors

A full list of current subprocessors is available upon request at [email protected]. We will notify you of material changes to our subprocessors.

17. Data retention

Waitlist data: we retain your name and email until you request deletion or until Pragmus launches publicly, at which point waitlist data is either migrated to your account or deleted.

Project content: retained for as long as your account is active. You can delete individual projects at any time if you are the project owner. On project deletion, all associated data is permanently removed — including submitted content, extracted items, generated summaries and structure, attached documents, and source records.

Account deletion: when you delete your account, your personal information (name, email) is removed. A pseudonymous user identifier is retained as a reference to contributions you made in shared projects. This identifier cannot be linked back to you.

18. Your rights

You can:

• Request access to the personal data we hold about you
• Request correction of inaccurate data
• Request deletion of your data
• Request a copy of your data in a portable format

To exercise any of these rights, email [email protected]. We will respond within 30 days.

19. GDPR — EU, EEA, and UK residents

If you are located in the EU, EEA, or UK, you have additional rights under the GDPR and UK GDPR, including the right to lodge a complaint with your local supervisory authority.

The legal basis for processing your data is:

• Contract performance — to provide the service you signed up for
• Legitimate interests — to operate and improve the service
• Consent — where required, such as for marketing communications

Our subprocessors operating in the United States rely on Standard Contractual Clauses or Data Privacy Framework certification for lawful data transfers from the EU. For details on specific providers, contact [email protected].

20. Security

We use industry-standard measures to protect your data, including encrypted connections (TLS), access controls, and isolated storage per account. No system is 100% secure. If you have security concerns, contact [email protected].

21. Children

Pragmus is not intended for use by anyone under the age of 16. We do not knowingly collect data from minors. If you believe we have collected data from a minor, contact us immediately.

22. Changes to this document

We may update this document from time to time. For material changes, we will make reasonable efforts to notify you via email. Continued use of the service after changes take effect constitutes acceptance of the updated terms and policy. The "last updated" date at the top of this page reflects the most recent revision.

23. Contact

For any questions, privacy requests, or legal notices: [email protected]