Privacy Policy

About this policy

This policy describes Robot Personal Assistant, a privately operated personal AI assistant built with OpenClaw, and this informational website. The assistant is used by its operator and people the operator expressly authorizes. There is no public account registration.

Visiting this website does not connect a Google account or give the website access to mail, files, calendars, or contacts. Google authorization happens separately through the operator’s existing assistant connections.

Google information and permissions

Robot accesses information from the Google account authorized by its operator. Google’s consent screen identifies the requested permissions. One connection is read-only; a separate automation connection has broader permissions for authorized tasks. A task may use only part of a granted permission.

Account identity

Email address and basic profile information to identify the connected account.

Gmail

Messages, attachments, labels, and supported mailbox settings. Authorized workflows may read, organize, draft, or send mail and perform supported mailbox changes.

Calendar

Calendars and events for schedule review, planning, and authorized creation, changes, or deletion of events.

Drive & Docs

Accessible files, document content, and metadata to find, summarize, and work with documents. The automation connection includes permissions to create and modify files.

Sheets

Spreadsheet content and metadata for retrieving information and maintaining personal trackers or other authorized spreadsheets.

Contacts

Contacts, other contacts, and directory information where available to the connected account, including authorized contact management.

These permissions do not make Google content public. Permission to access data is separate from authorization to perform a particular task.

How information is used

Robot uses relevant Google information to carry out the operator’s requested or previously authorized tasks: inbox review, scheduling, information retrieval, document and spreadsheet work, contact assistance, and delivery of task results.

Conversation context, saved task information, and search indexes can help the assistant retrieve relevant context for later requests. Automated outputs may be incorrect and remain subject to human review.

AI processing and sharing

Google data does not necessarily stay on the operator’s computer. Relevant messages, document excerpts, metadata, or derived information may be included in prompts and tool results processed by hosted AI services, currently OpenAI and Anthropic. Local models and local search indexes may also process information.

Hosted providers process submitted context under the terms and data controls of the product and account used. The operator has confirmed disabling optional general model-training and data-sharing settings in their OpenAI and Anthropic accounts. This is an operator-reported setting, not an independent audit of provider controls or internal practices.

Training opt-outs do not mean that data is never stored or reviewed. Providers may retain submitted context, process operational logs, or review information for safety, abuse prevention, support, or legal obligations under their applicable terms. Separate feedback or voluntary data-sharing features can have different rules; the operator does not intentionally submit Google content through those features. This policy does not promise zero retention or the absence of safety-related processing.

For authorized tasks, Robot may send results or selected information through iMessage or WhatsApp to intended recipients, write results to the connected Google services, or submit task-relevant information to remote APIs or search services. Those services and recipients may retain their own copies. Private Google content should be included only when needed for the authorized task.

The operator’s policy is not to sell Google data, use it for advertising, or use it to train a generalized AI model. Hosted AI processing must use service terms and account controls compatible with that policy and Google’s Limited Use requirements. Human access is limited to the operator, authorized recipients, or permitted security and legal purposes.

Provider information: OpenAI, Anthropic, Apple, and WhatsApp.

Puppy and voice information

Puppy is Robot’s experimental physical companion. Its voice service has not been approved for continuous operation. When its voice mode is running during supervised testing, its microphone processes nearby sound to detect a wake phrase locally. Nearby people’s speech can be picked up, and recognition can activate unintentionally. The operator should make people nearby aware of microphone use during supervised testing.

In the configured workflow, activation is followed by a short command recording, which is sent to the operator’s Robot computer for local Whisper transcription. The resulting text is passed to Robot. Transcripts and relevant task context, including Google information when needed for an authorized task, may then be processed by the hosted AI providers and connected services described above. Local wake-word detection and transcription do not mean the entire conversation stays on Puppy or the operator’s computer.

Spoken replies are generated locally and played through Puppy’s speaker, where other people nearby may hear them. Command recording, transcription, and speech playback use temporary audio files. Normal processing is designed to clean up those temporary files, but this is not a guarantee that no audio copies remain after an interruption or diagnostic session.

Recognized speech, requests, responses, and operational or sensor information can be recorded in logs. Task information may also remain in Robot’s conversations, files, and backups; diagnostic audio may be retained separately. The storage, retention, deletion, and provider limitations in this policy apply to those records. This website does not access a microphone or provide remote access to Puppy.

Storage and security

OAuth credentials are stored in the operator’s private credential storage, including macOS Keychain and access-restricted files. They are used to authenticate to Google and are not embedded in this website.

Google content or derived information may be retained in conversations, local task files, downloaded attachments, memory or search indexes, operational logs, and backups. Copies can also exist in the Google account, authorized messaging services, recipients’ devices, and hosted AI services.

The assistant uses authenticated Google connections and encrypted network transport for Google API requests. Local access controls protect its credential stores. No system can guarantee complete security; this policy does not claim that every local file or backup is encrypted or that all information remains on one device.

Retention and deletion

There is no single automatic deletion period covering all of Robot’s records. Task data and conversations may remain until the operator removes them or an applicable retention process deletes them. Backup copies follow the backup system’s retention cycle. External services apply their own applicable retention rules.

To request access to or deletion of retained task information, contact the operator using the support address shown on Robot’s Google consent screen or the private communication channel through which access to Robot was arranged. Identify the information or task concerned; do not send passwords or authorization tokens.

Deletion requires checking relevant conversations, files, indexes, logs, and backups separately. Revoking Google access does not automatically delete those copies. Deletion of a local copy also does not erase information already delivered to an authorized recipient or held by another service. The operator will explain any applicable retention limits rather than promise immediate deletion from every system.

Your control over access

You can review or revoke the application’s Google access in Google Account connections. Revocation prevents further access through the revoked authorization; it does not erase data already retrieved.

Ask the operator to stop a workflow or remove retained information through the same private support channel. Any new use of Google information requires an updated disclosure and consent when required.

This website

This site consists of static informational pages. Its files contain no analytics, advertising, tracking scripts, cookies, forms, browser storage, external fonts, or embedded third-party content. It does not receive Google OAuth credentials or connect to the assistant’s private runtime.

The public website uses Cloudflare Pages for hosting. Cloudflare and network providers process request information such as IP address, requested page, browser details, and connection timing to deliver and protect the site. Those infrastructure records follow the Cloudflare Privacy Policy and applicable provider practices. Website request metadata is separate from the Google data processed by the private assistant. Following an external link takes you to another provider’s website and its own privacy terms.

Google policy and updates

Robot’s use and transfer of information received from Google APIs will comply with the Google API Services User Data Policy, including its Limited Use requirements, and the Google Workspace API User Data and Developer Policy.

This policy will be updated when data practices change. The operator must obtain further consent where required before using Google information in a new way. Privacy questions and requests can be directed to the support address shown on the Google consent screen or the operator’s existing private support channel.