What Zero Retention Means
Zero retention is a design objective for Customer Content on systems that Mparanza controls, not a claim that every hosted service stores nothing. Customer Content means files, datasets, prompts, instructions, recordings, transcripts, reports, comments, and workflow outputs submitted for processing.
Ordinary plugin functions use your existing ChatGPT plan and Codex workspace without sending Customer Content to Mparanza. Mparanza-hosted services receive the content described for each service and follow the retention and deletion arrangements below.
Authentication, delivery, security, legal, saved-service, and external-service records are also addressed below.
Scope
This policy applies to mparanza.com, Mparanza-hosted tools, downloadable plugins, authentication and download flows, support communications, and other Mparanza services that link to it. It explains both retention and the wider handling of personal information.
Mparanza-controlled systems are systems that Mparanza operates or can directly administer. Your own device, existing ChatGPT plan and Codex workspace, connected services, email provider, and other externally operated systems are not Mparanza-controlled systems.
If you use Mparanza for an organization, you are responsible for giving required notices and obtaining the rights and permissions needed for information you choose to process.
Ordinary Plugin Functions
Ordinary plugin functions use your existing ChatGPT plan and Codex workspace. Files, scripts, and outputs may stay on your computer, while content Codex reads can enter the model context under that plan. Mparanza does not receive, access, or retain that Customer Content merely because you use the plugin, and the plugin does not automatically anonymize it.
Your ChatGPT plan, Codex workspace settings, connectors, and other services process information under their own arrangements. If a plugin clearly invokes a Mparanza-hosted service, that category is covered by the next section.
Vera and Clara work in ChatGPT with material supplied in the conversation and with callable connected apps. They may recommend Codex Desktop for direct folder access, persistent project files, local tools, and durable deliverables after providing useful work. Installation is optional and the conversation can continue in ChatGPT.
Vera's Studio Archive uses two non-hosted message routes. Gmail is searched from ChatGPT or Codex through OpenAI's separately connected Gmail connector. WhatsApp is inspected only from Codex Desktop with Computer Use in the WhatsApp Desktop app already opened and authenticated by the professional. Neither route creates a Gmail or WhatsApp message store on Mparanza-controlled systems. Screen text and images read by Codex may still enter the model context under the user's ChatGPT/Codex account.
Mparanza-Hosted Services
When you explicitly use a Mparanza-hosted service, the content needed for that service reaches Mparanza-controlled systems. Hosted processing may create uploads, temporary files, session or job state, extracted content, generated outputs, recordings, transcripts, or review artifacts.
Some hosted services process a request without intentionally preserving durable feature state. Others intentionally save a deck, template, checking session, interview, report, or output so it can be resumed, reviewed, shared, retried, or retrieved. Those saved records are exceptions to the zero-retention objective until they are deleted.
Closing a browser, completing a job, archiving a project, or allowing a link to expire does not necessarily delete the underlying hosted data. Mparanza's product development is focused on ordinary plugin functions and reducing these hosted retention exceptions.
Limited Information We Retain
Depending on how you use Mparanza, we may retain:
- account and authentication information, such as your email address and profile details used to sign in;
- Customer Content and feature state that you explicitly submit to or create with a Mparanza-hosted service;
- limited download, transactional-message, and delivery records, including email addresses or protected email identifiers and relevant timestamps;
- communications, feedback, support requests, and other messages you send to us; and
- technical and security information such as IP address, browser information, request path, timestamps, cookie or session identifiers, error logs, and security events.
How We Use Information
We use retained information for:
- to perform the workflow or provide the feature you requested;
- to authenticate users, secure sessions, prevent abuse, and protect the Service;
- to deliver requested downloads and transactional messages;
- to troubleshoot failures and provide support; and
- to comply with law, enforce terms, and protect rights, safety, and security.
Uses We Reject
Mparanza does not sell personal information, share it for cross-context behavioral advertising, or use Customer Content to train a generalized AI model. We do not claim ownership of Customer Content.
Mparanza will not use Customer Content for a materially different purpose unless that purpose is clearly disclosed and separately authorized where required.
Retention Today
Retention differs by service. The following describes the current behavior:
- Ordinary plugin functions: Mparanza retention is zero for content Mparanza does not receive. Content that Codex reads is handled under the terms and data controls of your existing ChatGPT plan and Codex workspace; local storage is not anonymization.
- Plugin updates and feedback: startup checks request the public plugin-version manifest and may poll the status of a previously submitted request. Those checks contain no Customer Content, although the technical records described below may be logged. Feedback or suggestion content is transmitted only through the explicit submission workflow and remains a support record until administrative deletion.
- Request-scoped hosted tools: the application does not intentionally preserve a durable feature copy after the response, although temporary system handling and technical logs may occur.
- Check Entries: server working files are ordinarily removed through periodic cleanup after about seven days; some job metadata uses a shorter, event-triggered cleanup. Cleanup is not guaranteed to occur at an exact instant.
- Hosted Voice call capture: live video remains in the browser and is not uploaded to Mparanza. Uploaded audio, upload chunks, and transcription work files are deleted before a completed package can be returned. On terminal package retrieval, transcript and package job state are scrubbed before the server sends the response; if that scrub fails, retrieval is blocked. When an authenticated launch includes compact case context, it is held in owner-bound opaque-token metadata and used to guide transcription. Token access expires after eight hours; startup and periodic cleanup remove expired launch metadata and abandoned call-capture state after interrupted processing.
- Retail data bridge: Retailer Signals and Brand Fit evidence jobs expire after 30 days, mapping worksets after 7 days, and mapping submissions after 180 days. Product-image bytes and generated reports are not uploaded by the plugin. These periods cover bridge artifacts; central structured product records, taxonomy, and accepted mappings are durable service data and do not share that automatic schedule.
- Saved Mparanza-hosted content: slide projects, templates, and hosted interviews are retained until manual or administrative deletion. Archiving a deck or expiration of an interview link is not deletion.
- Authentication: signed sessions normally expire after 12 hours. Sign-in links are single-use and normally expire after 15 minutes, although related delivery and technical records may remain separately.
- Technical records: web access logs currently use 14 daily rotations. Other application logs, download records, message records, caches, exports, and backups do not yet share one automatic deletion schedule and remain until rotation, cleanup, or administrative deletion.
- Diagnostic recording: where replay or diagnostic recording is enabled for a hosted workflow, prompts and outputs may remain until administrative deletion. Such a workflow is not a zero-retention workflow.
External Services
Ordinary plugin functions use your existing ChatGPT plan and Codex workspace. Mparanza does not control that plan or workspace or make promises about its handling of data; its terms and data controls apply separately.
When a Mparanza-hosted service uses an external service, the content described for that hosted service may be transmitted to it. External systems are not Mparanza-controlled systems, and their terms apply separately.
For Studio Archive, OpenAI's Gmail connector accesses the mailbox selected by the user in ChatGPT or Codex, while Codex Desktop Computer Use can inspect the WhatsApp Desktop interface on the user's own computer. The Gmail connector route can run in ChatGPT or Codex when connected; the WhatsApp route runs only in Codex Desktop. Gmail, WhatsApp, and OpenAI are external systems under their own terms and controls. The professional signs in to those services directly; Vera must not request passwords, QR codes, authentication cookies, tokens, or one-time codes through chat.
When Information May Be Shared
Mparanza may share information only:
- with service providers and contractors as needed to perform the purposes described in this policy;
- with your organization or workspace administrator when you use an organization-controlled account or project;
- to comply with law, legal process, or an enforceable request from a public authority;
- to protect rights, property, security, or safety;
- in connection with a corporate transaction; or
- with your consent or at your direction.
Deletion Requests and Privacy Rights
You may ask Mparanza to identify, access, correct, or delete information associated with you. We may verify your identity and ask for the feature, approximate date, or identifier needed to locate the information.
A deletion request covers Mparanza-controlled systems. It does not remotely delete files in your workspace, messages already delivered to recipients, recipient copies, or information held in systems that Mparanza does not control. Limited information may also be retained where required for law, security, fraud prevention, or the protection of another person's rights.
Depending on where you live, you may also have rights to portability, restriction, objection, withdrawal of consent, or appeal. Users in the EEA, UK, or Switzerland may complain to their local data-protection authority. Applicable legal bases may include performance of a contract, legitimate interests, consent, and compliance with legal obligations.
For California residents, Mparanza does not sell personal information or share it for cross-context behavioral advertising.
Security and International Processing
We use administrative, technical, and organizational measures designed to protect information. No online service, transmission, storage system, or model-based workflow can be guaranteed to be fully secure. Keep independent copies of important content and secure your account and devices.
Mparanza LLC is based in the United States. Hosted and external-service processing may occur in the United States and other countries whose data-protection laws differ from those where you live.
Children
The Service is not directed to children under 18, and we do not knowingly collect personal information from children under 18. If you believe a child has provided personal information to us, contact us so we can take appropriate action.
Changes
We may update this Zero Retention Policy as our workflows and controls change. The effective date identifies the current version. An updated version is effective when posted unless it says otherwise, subject to applicable law.
Contact
Zero-retention, deletion, and privacy requests should be sent to fabio@mparanza.com.