Understand the work. Review the access.

MyDataWork gives practitioners a workspace for the context around analytical work: connected assets, relationships, use cases, stakeholders and recorded outcomes. Leaders can use that context to inform priorities. It complements existing analytical tools and governance; it does not execute a migration, clean-room join or agent workflow.

The catalog holds three kinds of information:

• Asset identity: names, paths, file types, tool classifications, sizes and dates.
• Extracted structure: table and column identifiers, sheet names, source references and supported dependency information.
• Business context: tags, notes, use cases, objectives, progress, recorded value, stakeholders and relationships you enter.

Supported source files or definitions are opened or fetched to extract structure. Their contents and business-data rows are not retained as catalog content. Metadata and authored text can still contain sensitive information.

Extraction varies by format. Examples include SQL table references; Excel sheet names; CSV column names; structural references in notebooks; Alteryx inputs, outputs and join keys; ThoughtSpot object and table references; Dataiku recipe and dataset references; and Looker dimensions, measures and table identifiers. Cataloging an asset does not guarantee complete lineage.

See Source access & the Windows Connector for where processing occurs and what each connector reads.

No. Local scanning opens supported files on your computer. Uploads and some cloud connectors receive or fetch source content on the service to extract structure. The catalog retains extracted metadata rather than a copy of that source content.

Catalog retention is different from temporary processing, operational logs and backups. If your review requires specific temporary-storage, logging or deletion guarantees, request the details for the connection path you plan to use at contact@mydatawork.com.

The workspace database is hosted on AWS RDS for PostgreSQL in us-east-1 (Northern Virginia, United States). It is encrypted at rest with AWS KMS and is not publicly accessible.

Automated database backups have a seven-day retention period and support point-in-time recovery. Database location does not describe every processing path: applicable AI features and optional integrations involve the services described below.

Cloud connections read catalog metadata or supported source artifacts to extract structure. This is different from running analytical queries over business rows. Use the minimum permissions required for the selected connector; the following list describes the supported connection paths:

The Windows Connector recognizes supported Excel, CSV, SQL, Python and notebook files, BI exports and other supported analytical artifacts. It extracts filesystem information and available structural references; the amount of structure depends on the format. A cataloged file is not a guarantee of complete lineage or cell-level analysis. See the app's Metadata section for the supported formats and extracted fields. Local extraction occurs on your computer; uploads and cloud-source processing occur on the service.

Minimum permissions: GitHub repo read; Databricks workspace read plus Unity Catalog read; dbt Cloud read on projects/runs/artifacts; Snowflake SELECT on INFORMATION_SCHEMA only; BigQuery Metadata Viewer; Redshift read-only on system catalogs; Power BI read-only admin (service principal, Scanner API); Tableau Metadata API (PAT); Sigma/Looker/ThoughtSpot read-only metadata API; Dataiku read-only DSS key. Cloud connections are optional and admin-configured.

The Windows Connector runs on the user's computer and scans the folders assigned to a workspace. It communicates with the MyDataWork service over HTTPS and exposes a local control endpoint on loopback port 27843 for browser controls. Supported releases protect control operations with a local token. The browser and local security settings must permit the connection; verify the publisher signature when installing.

• Browser and Connector: HTTPS access to app.mydatawork.com on port 443.
• Local browser controls: the Windows Connector exposes a loopback endpoint on port 27843, protected by a local control token. Browser and endpoint-security policies must permit this local connection.
• Optional connections: cloud-source processing requires service-side access to the chosen provider. Jira uses HTTPS to your configured Jira instance.

These are distinct paths. A loopback endpoint is not a public inbound service. Organizations may still need proxy, firewall or endpoint-policy review; contact us for the connection details relevant to your deployment.

Local folder assignments belong to the individual computer and workspace. Pairing another computer does not copy the first computer’s folder list or require revoking its valid credentials. The catalog can receive results from multiple authorized computers. Folder assignment isolation does not change workspace member or administrator access to cataloged metadata.

Not for every record. Folder paths can be compared with the current computer’s scan folders, but that comparison does not establish which computer originally imported a record. A notice that assets are outside this computer’s current scan folders describes current path coverage, not proven origin.

No. It asks the Windows Connector (on the user's own machine) to open the file in its default application — the same local action already on the Assets tab. Nothing is uploaded, and MyDataWork gains no new data access. Cloud assets have no local file and deep-link in-app. The Workspace view is read-only over the existing catalog.

No. The connector is signed by MyDataWork, LLC, which identifies the publisher and allows signature verification. Browser download checks and Windows SmartScreen are separate checks, and results can vary by build, device and security policy. Setup provides download help and the published SHA-256 fingerprint. A valid signature or matching fingerprint is not a guarantee that a file is harmless; unexpected warnings should be checked rather than bypassed automatically.

Applicable AI features send relevant, authorized asset metadata and authored business context to OpenAI. This can include use-case titles, descriptions, objectives and structural observations.

Dedicated user and stakeholder name and email fields are excluded from model requests. This filtering applies to those specific fields; it does not automatically anonymize personal information entered elsewhere.

Depending on the feature, context can include titles, descriptions, asset paths, chat messages, communication audience lists and notes, and stakeholder roles. Names or other personal information entered in these fields can reach the model. Include only the information needed for the task; AI recommendations do not independently validate it.

Controls are in Setup → AI configuration. The app shows applicable allowances and costs. The Workspace Agent runs when requested; reading existing findings does not start a new analysis. Solution Pack starter content is not an AI assessment of your workspace.

The Asset Estate Assessment is an on-demand AI review using OpenAI. Its first run is free on each plan; subsequent runs use credits.

• Input: relevant asset metadata, use-case text and structural observations. Dedicated user and stakeholder name and email fields are excluded from model requests; titles, descriptions and objectives are not automatically anonymized.
• Grounding: findings are filtered against recorded workspace entities to reduce unsupported references. This does not guarantee accuracy or validate the underlying work.
• Storage: the rendered report is stored for the workspace; intermediate model outputs are not retained.
• Eligibility: a hashed-email record tracks free-assessment eligibility across workspaces. See the Privacy Policy for retained records.

Disabling AI hides the Assessment. Review findings before acting.

Acceptance records a proposed enrichment and its evidence in the catalog. It does not download and join the suggested dataset to your business data, execute a connection recipe or validate an improved outcome.

Suggestions use metadata and relevant authored context; AI-assisted suggestions can send that context to the AI provider. Public-source previews can fetch public data. Administrators choose permitted public sources in Setup → Public data sources.

Check the suggested source, connection keys, availability, licensing and suitability before using it in your own environment. A match based on declared metadata is not proof of an implemented data flow.

Agent Access optionally exposes selected workspace context to an external agent over the Model Context Protocol (MCP).

• Scoped grants: a user grants specific use cases to a specific key. The key can read only its granted context.
• Content: selected metadata, relationships and authored descriptions and objectives. It does not supply source credentials, source file contents or values from connected sources.
• Activity: key creation, grant changes and reads are logged. The owner can review Agent activity; members cannot inspect another member’s key labels, grants or tool arguments through that panel.
• Revocation: keys can be revoked. Deleted keys retain their activity history.

Keys are bearer credentials, stored hashed and presented over TLS. Live access is available on paid plans with a daily read allowance. Export provides a point-in-time context file on all plans; Explorer offers Export only. These allowances are separate from AI credits.

Review metadata and free text before sharing. An external agent’s access to other systems is governed separately. Revoking a key does not recall a file or context already supplied to an external recipient.

Agent Studio helps define an agent workflow before it is built. It does not execute the agent or its proposed actions.

• Discovery: reads advertised tool, resource and prompt names, descriptions and input schemas. Cataloging blocks invoking methods such as tools/call, resources/read, prompts/get and resources/subscribe.
• Catalog: records capability definitions and source information, such as server version, protocol, authentication mode and discovery timestamp—not the business results of executing tools.
• AI drafting: uses tool/source definitions and the design context you author. Review that text before requesting a draft.
• Handoff: an exported definition is a plan for implementation. Runtime access must be governed separately; discovery does not certify that a server is safe.

Team administrators manage members in Setup → Workspace management. They can create an account or link an existing one, subject to seat limits; existing credentials are preserved when linked. Members cannot self-join a workspace through this flow. Administrators can transfer ownership or remove membership; removal revokes access but retains contributed workspace records.

Ordinary members can see their own records and assets shared with them. Stakeholders are workspace business contacts. Administrators have broader access through the audited Workspace view; do not describe all workspace information as private to its author.

Selected assets can be shared and copied; copies are independent. Ordinary members do not directly share use-case records. Portfolio exports communicate the selected records within the caller's permitted scope. Initiatives and asset groups organize work but are not separate confidentiality boundaries. Administrators can inspect individual records through the audited Workspace view.

Architecture & Value is a deterministic leadership summary on eligible team plans. It uses recorded tools, use cases, value and utilization; it does not use AI or send this computation to an AI service.

Figures are recorded estimates and realized values, not independently verified financial results. The view does not ingest or infer platform spend.

Aggregation is not an administrator privacy boundary. Workspace administrators can inspect individual member records through the audited Workspace view. Ordinary members do not gain that access. Agree what belongs in the workspace before inviting colleagues or collaborators from another organization.

Jira is optional and configured by an administrator in Setup → Integrations. Its connection credentials are encrypted at rest and used to authenticate requests to your Jira instance.

Sending a use-case summary can transmit titles, descriptions, objectives, progress notes, stakeholder names and asset names. That content may be sensitive even though source file contents and underlying business-data rows are not included.

Review the summary and destination before sending. The integration is disabled by default and subject to rate limits.

Personal information can appear in several places:

• Account and member records: names and email addresses used for sign-in and workspace membership.
• Stakeholder records: names, optional email addresses, roles and notes links you add.
• Metadata and free text: names or other sensitive details entered in paths, titles, descriptions, objectives or notes.

AI-assisted features exclude dedicated user and stakeholder name and email fields from model requests. Names or other personal information entered in metadata or free-text fields are not automatically anonymized. Include only the personal information needed for your work, and do not enter credentials. See AI context sharing for details.

Limited eligibility and deletion-audit records use hashed identifiers. Explorer trial history can remain after account deletion. See the Privacy Policy for retained records and processing terms. MyDataWork does not sell data or use it for advertising; DPAs are available for organizational customers.

Passwords are never stored in plaintext — they're hashed with industry-standard PBKDF2-SHA256. Authentication uses session tokens valid for 7 days. Permissions are re-read from the database on every request, so changes to a user’s role or administrative access take effect on their next request. A permission change does not itself terminate the session; the token remains valid until it expires. Users reset passwords via email verification. We don't store payment card information; Stripe processes payments.

Services include AWS for hosting and transactional email, Stripe for direct payments, OpenAI for applicable AI features, and Jira when you enable that integration. Connected source providers are involved when you authorize those connections.

Review the Privacy Policy for processing terms. Contact us for a DPA or provider-specific information required for your review.

Workspace JSON: administrators can export from Setup → Data portability. The export includes assets and relationships, stakeholders, use cases and their recorded context, saved recommendations, action plans, initiatives, asset groups, function tags and agent designs.

It is not a complete system backup: it does not include source files, connector/cloud credentials or configuration, audit history, Assessment reports or Workspace Agent finding history.

Import requires administrator access to an eligible active workspace and is subject to plan limits. It merges supported records, adding or updating them rather than deleting destination records. Updates can change existing records; review the destination before importing.

Portfolio exports: PDF and PowerPoint presentations communicate selected work within the caller’s permitted scope. Exports remain available during the Explorer read-only window before deletion.

Keep workspace JSON exports as an independent copy of the records covered by the export. An import merges those records; it is not a full restoration of credentials, configuration, audit logs or all application state.

MyDataWork also maintains automated database backups with seven-day retention and point-in-time recovery. These operational backups are separate from user exports and are not a self-service restore feature. Contact us about recovery requirements for your security review.

Removal affects the catalog record, not the underlying file or cloud object. Removed records are hidden from ordinary catalog, use-case and lineage views and can be restored through the Removed filter. Stored use-case links survive removal. A scan may refresh metadata for a removed record but does not restore it. To stop further scanning of a local folder, unassign it with Stop scanning in Setup; that action does not remove its existing records.

Authorized users can select individual records or records matching the current filter. The action and confirmation show the affected count. Bulk removal and restoration are recorded as workspace events with the count and selector used. Requests remain subject to authorization and workspace scope; a matching path in another workspace does not grant access to that workspace.

Yes, through Account. Self-service deletion removes the account from the live service. Contributions in a shared Team workspace remain with that workspace; deleting an entire workspace removes its live content. Automated database backups have a seven-day retention period, and deletion-audit records use hashed identifiers. Records of lifecycle emails note whether a message was accepted by our mail provider; acceptance is not confirmation of delivery to the recipient.

Explorer is limited to one trial per email address. The eligibility record retained after account deletion contains a one-way hash of the address and the date it was claimed, not the original email address. This record enforces the one-trial limit and helps prevent abuse; deleting your account does not make you eligible for another Explorer trial. See our Privacy Policy. Deleting an account does not cancel an AWS Marketplace subscription.

Export your work while access is available. Subscription cancellation, account deletion and removal of catalog assets are different actions.

Explorer: after the 90-day trial, the workspace is read-only for 30 days. During that window you can export or subscribe. The trial workspace content is then scheduled for deletion; subscribing after deletion does not restore it.

For other subscriptions, review the applicable terms and Privacy Policy. Manage AWS Marketplace cancellation separately in AWS. Uninstalling the Windows Connector does not itself cancel a subscription or delete the workspace.

MyDataWork does not currently have its own SOC 2 report. AWS infrastructure assurance does not constitute a SOC 2 attestation for the MyDataWork application.

Controls described on this page include database encryption at rest, HTTPS transport, a private database, application network restrictions, backups and monitoring. Contact us for supporting information and the current assurance roadmap.

MyDataWork provides workspace export and self-service deletion features to support data portability and erasure requests. Account deletion, shared-workspace contributions and retained records have different treatment, described in our Privacy Policy. We do not sell data or use it for advertising. DPAs are available for organizational customers.

The core data-handling and security controls do not change simply because you choose Enterprise terms. Plans do differ in available features, capacity and access modes—for example, live Agent Access and team administration. Review the features you intend to enable as part of your assessment.

Yes. The public MyDataWork for Teams offer is a fixed 20-user annual package, purchased through AWS Marketplace and billed through AWS. Contact MyDataWork about larger teams or private offers.

Your organization’s procurement and security requirements still apply. See Pricing for the current package and trial terms.

The purchase channel does not change the application’s hosting region or core data-handling controls. The Marketplace package has its own capacities and commercial terms.

AWS manages the subscription, renewal and billing; MyDataWork checks entitlements and manages workspace members and roles. The public offer is a fixed package, not a self-service seat-adjustment plan.

Marketplace handles subscription and billing information: the purchasing account, entitlements, subscription status and charges. Seller reports identify the purchasing account.

Your catalog and authored workspace content are not supplied as Marketplace billing data. AWS also hosts MyDataWork independently of how you purchase, as described in the Privacy Policy. These hosting and billing roles should be considered separately.

Email contact@mydatawork.com. We aim to respond to security-related inquiries within 2 business days.