Frequently asked
questions.

XOS1 stands for XiteAI Operating System One. It is the AI-native operating system XiteAI is developing around conversation, memory and actions on your computer.

XOS1 is being developed for Windows PCs. Check the compatibility and installation requirements for the release you intend to use. Support for other platforms has not been announced.

XOS1 will launch in beta in India and the United States. Other countries are not included in the initial launch.

No public launch date is promised in the Beta Terms. Release information and availability will be announced through XiteAI’s official product and News pages.

XOS1 is designed for people of all ages, with an age-appropriate experience.

Under 13: Age-appropriate AI with respectful boundaries. The intended design includes optional parental controls; when enabled, they are intended to give a parent or guardian visibility into the child’s XOS1 activity.

13–17: Age-appropriate AI with respectful boundaries. The intended design does not include parental activity-monitoring controls for this age band.

18 and over: The full adult configuration, with its broader execution capabilities. The transition is intended to happen automatically when the recorded age reaches 18.

Age-specific model boundaries, parental controls and automatic adult-mode switching are part of the intended product design. Their implementation has not yet been verified in a customer release.

Lumen Visuals is a feature of XOS1. In XOS1, answers don't only speak. They appear around you.

Check the release information to see which features are available in your build.

XTEC is the XiteAI Terminal Enterprise Console. It is the workspace for the people behind XiteAI, with company and product tools available according to each person’s access.

Team sign-in is separate from browsing XOS1 or submitting an application for an open role.

The current product is being developed to run on a Windows PC. Its name does not mean you need to erase Windows or replace your computer’s boot system. Follow the installation instructions for the actual release.

XOS1 is being built around conversation, persistent memory and tools that work with information on your computer. The aim is to continue from what you have already shared and help with tasks, rather than only generate a text reply. The tools available depend on the release.

XOS1 is being developed for Windows PCs, but a verified minimum CPU, RAM, storage and GPU specification has not been published here. Check the requirements for your release before downloading or buying hardware. We do not promise that every Windows laptop is supported.

XOS1’s development experience includes spoken conversation and a text chat interface. Using voice and allowing biometric recognition are separate choices. Check the release’s input options and permissions before enabling them.

A confirmed language and accent support list has not been announced here. Do not assume that a model’s general language abilities mean every voice, transcription and product feature supports that language. Release information should identify the supported options.

The website includes illustrative conversations, visual scenes and examples of the intended experience. A flight animation or booking example is not proof that a live booking integration is available. Use the release information to check what your build can actually do.

No computer has literally infinite storage. In XOS1, “infinite memory” describes the aim of retaining useful memories across conversations, instead of starting over each time. Capacity depends on available device storage, and memory extraction and recall can still make mistakes.

We mean a continuing record that can grow with your conversations and provide relevant details later. It is not a promise of unlimited disk space, perfect recall, or every past conversation fitting into one AI request. Local stored memory and an AI model’s context window are different things.

The development design identifies useful facts and experiences from conversations, organises them into local memory records and retrieves relevant records for later questions. Dates and the people involved help distinguish one event or fact from another.

Relevant memory text can be included in a cloud AI request. Storing a memory locally does not mean every stage of understanding or answering it runs offline.

The current design groups memories into personal details, habits, skills, relationships, media and experiences. For example, a move to a new city, a skill you are learning, or a story you shared can help a later conversation. This describes the design, not a guarantee that every detail is captured correctly.

Not as a guarantee. Conversation history and selected memory records serve different purposes. The memory design aims to keep useful facts and experiences rather than treat every reaction or piece of small talk as a lasting fact. Extraction can miss information or interpret it incorrectly.

Yes. It can mishear a statement, confuse who it concerns, miss a correction or retrieve the wrong event. Check important dates and facts against the original record. Memory should help a conversation, not replace your own evidence or judgement.

No. Stored memory is information kept for later retrieval. A context window is the limited amount of information a model can work with in a particular request. XOS1 can select relevant stored records for that request; it does not send an unlimited lifetime of information into one prompt.

AI requests can include relevant memory and other context. The reviewed design is not a continuous upload of the whole local memory database, but the amount and type of context depend on the feature and request. Do not assume an answer uses only your latest sentence.

The memory design is intended to recognise clear corrections and update what is current while preserving the history needed to understand older conversations. Corrections can be missed, so verify important changes and use the available record controls in your release.

It should not. XOS1’s memory design distinguishes a current fact from something that was true earlier. A move or a new job can update the present while earlier conversations keep their historical meaning. That process is not error-free.

The memory design keeps dates with records so a past detail can be connected to the conversation or event it came from. Retrieval depends on what was actually saved and how accurately it was interpreted; it cannot reconstruct an event that was never recorded.

Reviewing, correcting and deleting memory records is part of the intended product experience. Check the controls in your release for the exact scope. Local factory reset clears memory within its reset scope; a company-held data request is a separate process.

The design ties relationship details to the person they concern and aims to ask when a name is ambiguous. It can still make attribution mistakes. Do not treat another person’s statement as proof about you, and only share their information where you have the right to do so.

A verified automatic memory-sync or migration feature has not been announced here. Records described as local live on the device where they are stored. Do not assume signing in elsewhere transfers them; check the release’s supported backup and migration options.

Local memory needs disk space. Low storage can affect saving, updates and other app functions. The term “infinite memory” does not bypass that limit or guarantee data recovery. Keep room available and back up important information.

Uninstalling an app is not proof that every local record or backup has been erased. XOS1’s factory-reset function and a company-held deletion request have different scopes. Neither should be treated as a secure wipe of exports, backups or your whole disk.

XOS1 is being built to do more than answer questions: it can work with supported tools and files on a Windows PC. “Computer use” is not a promise that it can operate every app, website or device. Check which actions and integrations are supported by your release.

The reviewed design includes an XOS1 workspace and read-only browsing of the wider computer. Relevant text from files, including an open workspace document, can enter AI context. Pointing XOS1 at a file can therefore involve external AI processing, not just local storage.

In the reviewed design, XOS1 can read and write in its own workspace. Writing or deleting outside that workspace is restricted by default; broader device write access is a deliberate setting. Permissions reduce scope, but they do not eliminate mistakes. Keep backups.

This FAQ does not claim continuous screen monitoring. File context, camera recognition and a particular computer-use feature are different sources of information. Check the scope of any screen or app access in your release rather than assuming it sees everything, or that it can never access visible information.

The documented design includes approval for destructive actions. Review what the proposed action will affect before approving it. An approval prompt is a safeguard, not proof that the action is harmless or that all possible errors have been prevented.

Only enable broader access when you understand why the task needs it and which files may be affected. The default workspace restriction is narrower. A useful AI response does not give the system authority over every file or account on your computer.

That depends on a supported, authorised integration in the release. Website examples are not confirmation of live purchasing or messaging support. Review recipients, content, price and the final action before approving anything with consequences; a subscription payment does not authorise unrelated purchases.

Check the actual result. An AI reply can be wrong, and a tool or outside service can fail. A generated message is not a receipt, a booking confirmation or evidence that a file operation completed.

Avoid putting secrets into general conversation. Context can be processed by a cloud AI provider. A specific protected-file or account flow must explain how a secret is handled; do not assume any password entered anywhere is automatically excluded from history, memory and external processing.

Do not assume that it will. Background activity, reminders and autonomous execution depend on the running components and features available in your release. A website demonstration is not a promise that an unsupervised task will continue after you close the app.

No. In the reviewed build, records such as conversation history and memories are stored locally, but AI request text and relevant context can pass through XiteAI’s gateway to the configured AI provider.

The reviewed gateway relays requests without keeping a company archive of their contents. Local storage, external processing and company-held service records are different things. The Privacy Policy explains each.

No. Local storage describes where records are kept. AI requests, web information and downloads can need an internet connection or an external service. Do not assume a feature works offline because its records are local.

Deleting records on your device and asking XiteAI to delete company-held information are separate actions. The Data Deletion page explains both and how to make a request.

A local factory reset is not a secure wipe of your disk or external backups. Clearing local records does not recall information already processed by an external service.

Yes. Open Privacy choices to allow or refuse optional interface settings. Necessary storage supports services you request, such as sign-in and application drafts.

Refusing optional settings does not prevent you from browsing, applying or signing in. These website choices are separate from settings inside the XOS1 desktop application.

Use the Security page or the Security contact desk. Describe the affected feature and what you observed. Do not send passwords or unrelated personal information, access someone else’s data or disrupt a live service to demonstrate a problem.

XOS1 is designed to keep voice transcription, biometric templates and many personal records on your device. Cloud AI requests and service records still involve data leaving the device. Privacy depends on those data flows and the controls actually present in your release, not on calling the product private.

The reviewed basic check-in includes an install identifier, signing public key, app/build and operating-system versions, device type, language, time zone, update status and crash counts. Network information can also be processed for security and service operation.

Optional profile sharing can send a name and date of birth; optional usage sharing can send tool names. Both default to off. Support messages, submitted applications and company-account records are separate information you provide or generate through those services.

Conversation text and relevant context can pass through XiteAI’s gateway to the configured AI provider. The reviewed gateway does not keep a company conversation-log archive. Passing through a service for processing is different from storing a permanent archive.

XiteAI’s policy is not to use private customer content for model training or authorise its providers to do so. Final production provider agreements and account controls still need verification. A no-training commitment and a verified provider configuration are different things.

The Privacy Policy states that XiteAI does not sell personal data or share it for advertising. This does not mean no information leaves your device: AI processing, requested external lookups and service operation have their own disclosed purposes.

DeepInfra is used in development testing. Cerebras and Groq are being evaluated for the beta launch; they are not confirmed production recipients here. The customer release must identify the actual providers and safeguards before enabling those cloud features.

No. XiteAI’s planned production requirement is zero retention of customer prompts and responses, but this is not yet verified across all selected provider accounts and features. Account, billing, security and usage metadata are separate from conversation content and can have different retention rules.

In the reviewed desktop build, speech transcription runs locally and raw microphone audio is not sent through the AI gateway. Transcribed text and relevant context can leave the device. A locally produced interpretation of voice state can also enter AI context.

No. The camera is off by default in the reviewed design and is not opened or sampled while it is off. Enabling a camera feature is separate from accepting general terms. Check what the feature will do and what permissions it needs before switching it on.

Biometric recognition is intended to be optional, with separate enrolment, disable and deletion choices. The reviewed development onboarding currently learns a voiceprint without a separate opt-in, so those planned customer controls still need implementation and verification. Voice transcription and creating a biometric identifier are different functions.

No system-wide keystroke recording is described for the reviewed Pulse feature. It is off by default and, when enabled, measures typing timings and counts in XOS1’s own chat box. A local typing profile and a summary sent with an AI request are different information.

It can estimate some indicators from interactions, such as energy or tiredness in speech. Those estimates can be wrong and are not a diagnosis or direct knowledge of your feelings. Separate Emotional Presence and Pulse features are built but off by default in the reviewed release.

The reviewed gateway does not create a company archive for staff to browse. Staff can receive the information you deliberately send in a support request, and cloud providers process the content involved in AI requests. We do not claim that no person could ever access data through a compromised device, service or account.

Do not assume that local storage means end-to-end encryption. The reviewed records rely on your operating system’s account and device security. Someone who can access that account or its files may be able to read them. Use a device password and appropriate disk encryption.

The reviewed XOS1 privacy settings separately control profile and tool-usage sharing, and both default to off. Turning profile sharing off clears the current shared name and date of birth at the next successful check-in. Turning usage sharing off stops future reports but does not automatically delete older service records.

The planned individual monthly prices, before applicable taxes, are:

Premium: $20 per month.

Ultra: $60 per month.

Max: $350 per month.

These are planned launch prices, not proof that checkout is live. The selected plan, included features and usage allowance must be shown before you pay.

Final trial eligibility, payment-card requirements and conversion terms have not been announced here. A trial recommendation in an internal draft is not a public offer. Only rely on the terms shown when a trial is actually available.

Do not assume that it does. Models, features and measured usage allowances for each tier are still release decisions. A plan must explain its actual limits and reset rules before you subscribe; “infinite memory” is not a promise of unlimited model requests.

A separate memory fee has not been announced here. Local capacity depends on your device storage, while AI processing and features depend on the plan’s published terms. Check the complete offer before paying rather than assuming every memory-related feature is included.

The planned billing currencies are USD and INR. Final INR prices have not been published here. The checkout must show the currency, billing period, applicable taxes and total before payment. The listed monthly USD prices are before tax.

The subscription terms describe automatic renewal until cancellation, with cancellation stopping the next renewal and access continuing through the paid period, subject to applicable rights. Check the available billing controls for your release; this FAQ does not claim an unverified checkout or cancellation flow is already live.

Voluntary cancellation or a change of mind does not ordinarily qualify for a refund, except where applicable law gives you that right. Verified duplicate or erroneous charges are refunded. Discontinuation of a paid service has its own elapsed-period refund schedule.

The planned launch subscriptions are for individual customers. Team and enterprise arrangements are planned separately; a household-sharing offer has not been announced. Paying for a child’s account does not by itself grant access to that child’s conversations.

Age-appropriate AI with respectful boundaries. The intended design includes optional parental controls; when enabled, they are intended to give a parent or guardian visibility into the child’s XOS1 activity.

Parental controls and parental consent are separate. Where applicable law requires parent or guardian consent for a child’s personal information, that consent must be obtained before the relevant processing. The absence of parental monitoring for an age band does not remove a legal consent requirement.

Age-specific model boundaries, parental controls and automatic adult-mode switching are part of the intended product design. Their implementation has not yet been verified in a customer release.

Age-appropriate AI with respectful boundaries. The intended design does not include parental activity-monitoring controls for this age band.

Separate guardian-consent or data-request rights can still apply under local law. The absence of a monitoring feature does not remove those requirements.

Age-specific model boundaries, parental controls and automatic adult-mode switching are part of the intended product design. Their implementation has not yet been verified in a customer release.

The intended design uses the recorded date of birth to switch automatically to the adult configuration when the user turns 18. That configuration continues to follow permissions, privacy settings and the Terms of Service.

An age transition must not automatically enable a camera, enrol biometrics, expand file permissions, turn on optional reporting, authorise a new payment or provide consent for a new use of personal information. A birthday does not automatically erase existing records; their retention and deletion remain subject to the Privacy Policy.

Age-specific model boundaries, parental controls and automatic adult-mode switching are part of the intended product design. Their implementation has not yet been verified in a customer release.

No. The intended under-18 experience uses respectful, age-appropriate boundaries, but AI can make mistakes and safeguards can fail. XOS1 is not a replacement for a parent, teacher or caregiver. Child-specific processing must also meet the applicable privacy and consent requirements.

XOS1 does not replace a qualified doctor, therapist, lawyer, financial adviser or accountant. Verify important information and use appropriate professional help for high-stakes decisions. It is not an emergency service.

Untrusted content can contain misleading instructions, and an AI system can misinterpret them. Do not grant access or approve an action just because a file or page tells you to. Review consequential actions and report suspicious behaviour. This FAQ does not claim that prompt-injection attacks have been eliminated.

Local data depends on the security of the device and account that hold it. A device password, appropriate disk encryption and secure backups help protect it. No remote-wipe or automatic recovery guarantee is made here. Contact support or the privacy team if you need help with company-held records.

Find a role on the Open roles page, read its description and select Apply. The application asks about your background and work, and lets you review your answers before submitting.

Use the role description for its specific responsibilities, eligibility and work arrangement.

No. Applications for publicly listed roles can be started without signing in to XTEC. Team access and applications are separate processes.

Application answers are saved on the browser and device you are using. Passwords, consent confirmations and skill-proof photos are excluded from that saved draft.

A draft is not a submitted application. Review your answers and submit when you are ready. Clearing browser storage can remove the draft.

The Careers page outlines application review, screening, a role-specific interview, and offer and joining. We discuss your work and the thinking behind it.

We do not use standard coding tests. If a practical exercise is part of the interview, it is open-ended; you may use the internet, and the time taken is recorded.

Immediate hire is for candidates invited directly by XiteAI. It uses the access password shared by your XiteAI contact. It is separate from applying for a publicly listed role.

The Company Policy page explains the existing XiteAI Code and identifies additional working practices that are still proposals.

Use the help desk to describe what happened and what you expected. Include your install code if you have it. You receive a reference that you can use to check the request.

For application questions, use the Careers contact desk instead. Privacy and security concerns have their own contact routes.

In XOS1, open Settings → About. The install code helps support identify the installation involved in your request. It is optional in the help form.

Use the request reference and the same email address you supplied when you contacted the help desk. The request checker shows its current recorded status.

Choose An idea in the help desk or use the Feedback contact desk. Tell us what you were trying to do and what would make the experience better.

Sending feedback is not permission for XiteAI to publish it as a testimonial. Public user stories will need permission from the people involved.

Use XiteAI’s official product and News pages. Posts marked preview or illustrative are demonstrations, not released product announcements. Use the actual release information for availability, supported features and requirements.

Include the version, operating system, the steps that caused the problem, and what you expected instead. Add a redacted screenshot if it helps. Keep passwords, API keys, payment credentials and unrelated private conversations out of the report.

Contact the team responsible for your question. Product help, privacy, security, billing and careers enquiries have separate routes on the Contact page. Tell us what you need to understand rather than sending unnecessary personal information.