Legal & Compliance

Terms of Service

Version: v2.5Effective date: 2026-10-23Last updated: 2026-10-23Last reviewed: 2026-10-23

What changed in version 2.5 (effective 2026-10-23). The paragraph of Section 2B on approvals now matches the AI Employees Notice, version 1.45: where the Customer leaves on the setting "Decides its team's requests" of an AI Executive, that AI Executive may decide a request to add a website to the websites an AI employee below it may reach — a website on Vezoft's list of common software sources, as until now, or, from the day Vezoft switches it on, a website that passes TimerOS's safety check, for an AI employee of its own line and where the request names an open task of that AI employee, as the Notice describes. A website it approves in this way goes on a list kept for the workspace and then opens at once for another AI employee of that line that asks for it, naming an open task of its own, where that AI Executive could decide the request itself; the person who answers for that AI employee's requests is told and can undo it within 7 days, and a person who holds the "Manage AI employees" permission can remove a website from the list, which closes it on every AI employee's computer. Section 2B's list of what the Customer decides now says this too. In version 2.4 a website not on Vezoft's list was always decided by a person. Because it widens the websites an AI employee may open for another AI employee, this is a material change under Section 15. It takes effect on publication rather than after 30 days' notice, because on 2026-10-23 the Service had no Customer other than Vezoft itself, and therefore no one who could be given notice or could object; account owners are asked to accept version 2.5 in the applications. Version 2.4 applied until 2026-10-23.

What changed in version 2.4 (effective 2026-10-13). The paragraph of Section 2B on approvals now matches the AI Employees Notice, version 1.21. An AI employee may ask for a website to be added to the websites it may reach. Where the Customer leaves on the setting "Decides its team's requests" of an AI Executive, that AI Executive may now also decide such a request for an AI employee below it, but only for a website on the list of common software sources that Vezoft keeps — package registries, official download and update sites, operating-system update servers, documentation sites and the code host's release downloads — and never for a file-sharing, paste, tunnelling or user-hosting site; any other website is decided, as before, by the first person above the AI employee. That person answers for the AI Executive's decision, sees it in the Service and in a daily digest, and can undo it within 7 days, which removes the website again. The paragraph also says that an AI employee may connect software it installed on its own computer to its work sessions as a local tool server, without anyone's approval and within the same limits as everything else on that computer, and Section 2B's list of what the Customer decides now says that an AI Executive may add websites in this way. In version 2.3 an AI Executive decided no website. Because it lets an AI employee decide which websites another AI employee may reach, this is a material change under Section 15. It takes effect on publication rather than after 30 days' notice, because on 2026-10-13 the Service had no Customer other than Vezoft itself, and therefore no one who could be given notice or could object; account owners are asked to accept version 2.4 in the applications. Version 2.3 applied until 2026-10-13.

What changed in version 2.3 (effective 2026-10-11). The paragraph of Section 2B on approvals now matches the AI Employees Notice, version 1.16. An AI employee's request no longer always goes to a person: where the Customer leaves on the setting "Decides its team's requests" of an AI Executive or a team lead — on the AI employee's own page; on by default for AI Executives and off for team leads — that AI employee may decide certain requests of the AI employees below it about AI employees' work on internal projects without a client and, for an AI Executive, also pushing code and having a pull request merged; otherwise the request goes, as before, to the first person above the AI employee. No AI employee decides its own request, a request filed by or about a person, payments, e-mail outside the workspace, publishing (though a merge may still deploy through the Customer's own pipeline), deleting records, its spending cap, anything about a public repository, or a push that changes a repository's CI or test configuration. A pull request is merged through the code host's auto-merge where the repository's branch rules apply; otherwise, on an AI employee's decision, only when the repository's branch rules require checks, each check they require succeeded on that exact commit (a skipped check does not count), every other check on it passed and the deciding AI employee states that it is confident, and in any other case a person decides. Every AI decision is carried out under the permissions of the first person above, who answers for it, sees it in the Service and in a daily digest, and can undo it within 7 days. A person's refusal sends that exact request back to a person for good, however often it is filed again; an undo, or a refusal where the person so chooses, also keeps the repository, project or AI team concerned — and the plans, work proposals and new-project requests of the AI employee that asked, and for a new project those of every AI employee — with a person for 7 days, unless a person in the line ends it sooner; and a merge is decided on the code host's own record of the pull request. The deciding AI employee is instructed to pass up to a person anything that looks off. A person may also approve pushes or merges in a repository in advance by a standing permission. Version 2.2 said that an AI team lead is passed over and never approves, and that an AI employee can never merge a pull request. Version 2.3 also adds obligations of the Customer as deployer: the branch rules and checks of its repositories, and keeping the TimerOS GitHub App off any list that bypasses them. Because it changes safeguards that version 2.2 described and adds obligations of the Customer, this is a material change under Section 15. It takes effect on publication rather than after 30 days' notice, because on 2026-10-11 the Service had no Customer other than Vezoft itself, and therefore no one who could be given notice or could object; account owners are asked to accept version 2.3 in the applications. Version 2.2 applied until 2026-10-11.

What changed in version 2.2 (effective 2026-10-10). The paragraph of Section 2B on approvals now matches the AI Employees Notice, version 1.14. The manager who approves is the first person above the AI employee in its reporting line; an AI team lead on that line is passed over and never approves. On its own virtual computer an AI employee installs software, runs commands, and edits and deletes files without anyone's approval; version 2.1 said that deleting data outside its own working files needed its manager's approval. Acts that reach outside its computer still need its manager's approval of the specific act, or are blocked; the list now also names pushing code and opening a pull request, and deploying, other than a website preview, which needs no approval; and it says that the e-mail concerned is e-mail outside the workspace. The AI employee files its request and carries on with other work instead of waiting for the answer; an approved act is carried out at most once, exactly as it was filed, within 7 days of the decision; and it can never merge a pull request. The paragraph also says that the request is filed on the AI employee's own computer and not checked by the Service's servers, and that the AI employee does not ask before sending data to a website its firewall admits, such as by submitting a web form. Nothing else changes. Because it removes approvals that version 2.1 required, this is a material change under Section 15. It takes effect on publication rather than after 30 days' notice, because on 2026-10-10 the Service had no Customer other than Vezoft itself, and therefore no one who could be given notice or could object; account owners are asked to accept version 2.2 in the applications. This is the first material change since version 2.0: the advance notice promised in the note on version 2.0 and in Section 15 is owed to Customers, and it applies in full from the first Customer on. Version 2.1 applied until 2026-10-10.

What changed in version 2.1 (effective 2026-09-16). One sentence in Section 2B is corrected. Model inference for AI employees now runs through a routing gateway that forwards each request to one of several model providers, so the Terms no longer say that it takes place in the United States specifically: they say that it takes place outside the European Union, and point to the Sub-processors page, which names the gateway, every model provider it may route to, where each one processes, and the safeguards. Nothing else changes — no right is reduced, no obligation added and no price altered — so under Section 15 this correction takes effect on posting rather than after 30 days' notice. Version 2.0 applied until 2026-09-16.

What changed in version 2.0 (effective 2026-09-15). A new Section 2B lets a Customer on the Startup or Enterprise plan hire AI employees that Vezoft operates on its behalf, and sets out the human-manager and approval rules, the fees (monthly rate, included active hours, overage, spending cap), the new sub-processors this involves (AI model providers in the United States and a hosting provider in Germany), and the absence of any warranty for their output. Sections 3 and 11 cross-reference it. Nothing else changes. Version 2.0 is in force now: it took effect on publication rather than after a notice period, because at that date the Service had no Customers bound by version 1.0 and therefore no one to give notice to. Every later material change is published with advance notice as described in these Terms.

These Terms of Service ("Terms") form a binding agreement between you and Vezoft EOOD ("Vezoft", "TimerOS", "we"), a single-member limited liability company incorporated in Bulgaria that operates the "TimerOS" service. See our Imprint for full company details. If you accept these Terms on behalf of an organisation, you confirm that you have authority to bind that organisation, and "you" / "Customer" refers to that organisation.

By creating an account or using the Service you accept these Terms. If you do not accept them, do not use the Service.

1. The Service

The Vezoft TimerOS Service (the "Service") is a multi-tenant enterprise time-management and operations platform currently delivered through the following coordinated surfaces:

  • (i) Windows desktop application — installed on the user's computer. Supported platform at launch: Microsoft Windows; native macOS support is planned. Consists of an always-on-top timer bar and a full-feature dashboard window. This is the only surface that performs automatic activity time tracking and runs the on-device activity classifier.
  • (ii) Web dashboard application — a browser-based version of the dashboard, mirroring the desktop dashboard's management and reporting functionality. Accessible from any modern browser at a tenant-specific URL. It does not track time or run the classifier.
  • (iii) Android companion application — a mobile review-and-management dashboard, distributed via Google Play, that mirrors the web/desktop dashboard's management and reporting functionality optimised for handheld use. It is a companion surface only: it does not track time and does not run the activity classifier (those remain Windows-desktop-only). Native iOS support is planned (not yet available).
  • (iv) Per-tenant, branded client portal — a separate web application that the tenant's external clients use under the tenant's own branding.
  • (v) Marketing and admin website at timeros.ai — the product marketing site and registration flow, hosting the owner/admin console for billing, subscription, users, and account-level settings. (Corporate/company information is published separately at vezoft.com.)

The set of surfaces, and the functionality available on each, may evolve over time as new platforms are added (e.g., macOS desktop, iOS companion app) and as existing surfaces gain new features. Any material expansion of the delivery surfaces, or any change in how personal data is processed, will trigger a corresponding update to these Terms (and, where appropriate, to the Privacy Policy, the Sub-processors list, and the Data Processing Agreement). Feature additions within an existing surface that do not change the categories of personal data processed do not require an update to these Terms but may be reflected in product release notes.

The dashboard (whether accessed via the desktop, web, or mobile surface) provides the day-to-day operational surface organised around Work, Organization, and Business areas, covering role-aware home dashboards, personal Schedule, Shifts (templates and assignments), Projects (with overview/team/budget/files/notes/activity tabs), cross-project Tasks, a Clients CRM, internal Support/incident tracking with SLA timers, Goals and Awards/Achievements, Team and Department management, an Organization chart with pending placements, Roles and permissions (presets and custom), an Approvals inbox, Billing/invoicing, Payroll (with Bulgarian-style deduction components and payslip generation), Expenses (bills, transactions, transfers, expense-to-invoice), Profit-and-loss analytics, and user account Settings. Capabilities are gated by subscription tier (Freelancer/Startup/Enterprise) through entitlements, and by employee role through a permissions layer.

The TimerOS Windows desktop application is REQUIRED for automatic activity time tracking — no other surface tracks activity. Reading the title of your current foreground window and the name of the running application, measuring the time since your last keyboard or mouse activity to detect idle and breaks, and the on-device activity classifier (a scored rule engine combined with a small on-device statistical classifier — no cloud AI service, no model download, no large language model) all run exclusively inside the Windows desktop application on your device. The web dashboard application, the Android companion application, the marketing and admin website, and the client portal can only display, manage, and report on data the desktop application has produced; they cannot capture activity data themselves. A user who does not install and run the Windows desktop application cannot generate productive or idle classifier outputs, cannot have shift-based hours classified, and cannot record per-window or per-application activity.

The marketing and admin website at timeros.ai hosts the owner/admin console for the SaaS account (workspace overview, Billing, Subscription, Users, Settings, Support tickets, in-app product Docs) and performs no activity tracking of its own. The client portal is a separate, per-tenant branded web application with its own authentication, giving the tenant's clients permission-scoped access to submit and track Projects, Incidents (with impact/urgency-derived priority and SLA tracking), ITIL-style Change Requests, view and pay Invoices via Stripe Checkout, manage Contacts and per-module permissions, maintain a Company Profile with 2FA, and review their Service Plan/SLA entitlements. The Service is described at https://timeros.ai.

2. Your account and authority

You must provide accurate registration information and keep it current. You are responsible for the security of your credentials and for all activity in your account. You must be at least 18 years old and capable of entering into a binding contract.

The Service is offered in two sales contexts: (a) the Freelancer plan, which is sold to individuals — including self-employed persons, freelancers, sole traders, independent contractors, and solo consultants — for use in their independent professional or business activity; and (b) the Startup and Enterprise plans, which are sold to legal entities (companies, partnerships, non-profits, public bodies) for use by their personnel. If you accept these Terms on behalf of an organisation under (b), you confirm that you have authority to bind that organisation, and "you" / "Customer" refers to that organisation. The Service is a professional time-management, billing, and operations tool; it is not designed for purely personal, household, or recreational use unrelated to a professional or business activity.

Where you qualify as a "consumer" under applicable law (for example, a natural person acting outside trade, business, craft, or profession within the meaning of Article 2(1) of EU Directive 2011/83/EU), the mandatory rights conferred on consumers — including the 14-day right of withdrawal where applicable — apply notwithstanding any other provision of these Terms and are not waived. The "authority to bind the organisation" language in these Terms applies only to sales context (b). The DPA applies in both sales contexts (see Section 5): in context (a) it governs personal data of individuals other than you — for example your clients and their contacts — that you process in the Service, and "Customer" in the DPA refers to you in your professional capacity. Obligations in these Terms and the DPA that concern a workforce — including the Employee Monitoring Notice obligations in Section 6 — apply only where and when you have workers other than yourself using, or tracked by, the Service.

2A. Connected AI assistants and third-party tools

The Service lets a signed-in user connect an AI assistant of the user's choice (for example Claude, made by Anthropic) to that user's own account through the TimerOS connector, a Model Context Protocol (MCP) server that we host. The following applies to every such connection:

  • The user's choice and authorisation. Connecting an assistant is optional and is the user's own decision. By completing the sign-in and consent steps, the user authorises the assistant to act in the user's name within the scope approved on the consent page (read only, or read and write, together with — where the user leaves the separate checkbox ticked — work on the workspace's incidents and change requests) and within the permissions that user already holds in the workspace. The assistant can never see or do more in the Service than the user could.
  • The assistant's actions are the user's actions. Every request the assistant makes and every change it writes is treated under these Terms as made by the user who connected it, exactly as if the user had made it in the desktop or web application — including for the purposes of Sections 4 (Acceptable use), 6A (Manual time-entry editing), and 13 (Indemnification). That includes assigning an incident or a change request to a colleague and, where the user approves a plan the assistant puts in front of them, resolving an incident or completing a change request; notes the assistant adds to such a ticket are internal to the Customer's staff and are not shown to the client. The Service journals each change the assistant writes together with the call that reverses it and shows the journal under Settings → Connections; reviewing that journal and reversing unwanted changes is the responsibility of the user and of the Customer.
  • Choosing and contracting with the provider. The user and, where applicable, the Customer are solely responsible for choosing the assistant provider; for reading, accepting, and complying with that provider's terms; for what the provider does with the data disclosed to it; and for ensuring that use of the feature complies with the Customer's own internal rules, confidentiality obligations, and any contractual or legal restrictions that apply to the Customer's data (including data of the Customer's clients). The provider is not our sub-processor (see Section 4A of the Privacy Policy), and we are not a party to any agreement between you and the provider.
  • Workspace control. A workspace owner or administrator may turn assistant connections off for the whole workspace in Settings → Connections; doing so disconnects every live connection in the workspace immediately and blocks new ones. Where the Customer is the controller of its workers' personal data, deciding whether its people may connect assistants — and under which internal rules — is the Customer's responsibility.
  • Suspension. We may at any time suspend or disconnect any connection, or suspend or withdraw the feature as a whole, where we reasonably consider it necessary for the security or integrity of the Service, to prevent abuse, to enforce these Terms or the Acceptable Use Policy, or to comply with law; we may also rate-limit connections separately from the user's own limits.
  • No warranty for the assistant. The assistant is a third-party product that we do not control. We give no warranty of any kind as to the assistant's behaviour, availability, accuracy, or output, or as to the correctness of any change it writes to the Service on the user's instruction; Section 11 applies. Any output the assistant produces from data obtained through the connector is the assistant's output, not ours.
  • Data roles unchanged. Nothing in this Section changes the allocation of controller and processor roles in Section 5 or in the DPA. Data disclosed to the assistant provider is disclosed by the user (and, where the Customer is the controller, on the Customer's responsibility) to a recipient of the user's choosing, not by us.
  • Other third-party tools. The same principles apply to any other third-party tool or client that a user connects to the Service through the connector. At the date above, the connector is the only integration the Service offers.

2B. AI employees

On the Startup and Enterprise plans the Service lets the Customer add one or more AI employees to its workspace: software agents, driven by third-party AI models, that Vezoft operates on the Customer's behalf on dedicated virtual computers, and that appear in the Customer's team with a name, a job title, a tier, a team, a human manager and working shifts. The following applies to every AI employee:

  • Beta. AI employees are offered as a beta: the feature is new, its capabilities, behaviour and pricing may change, parts of it may be reworked or withdrawn, availability is not guaranteed, and Vezoft gives no service-level commitment of any kind for AI employees.
  • What Vezoft provides. Vezoft provides and operates the agent software, one virtual computer per AI employee in the European Union, the connection to the AI model providers under Vezoft's own accounts, the approval, audit, recording and cost-tracking mechanisms described in the AI Employees Notice, and the controls in the workspace to pause, suspend and remove an AI employee. Vezoft is the provider of the AI system within the meaning of the EU AI Act; the Customer is its deployer.
  • What the Customer decides and is responsible for. The Customer decides whether to hire an AI employee, what it works on, which tools, websites and accounts it may use (an AI Executive may add websites from Vezoft's list of software sources or, since version 1.45 of the AI Employees Notice, websites that pass TimerOS's safety check, for its own line, as described below), who its manager is, and when it is paused or removed. The Customer is responsible for the instructions it gives an AI employee; for what it lets an AI employee read, including data of its own clients; for reviewing what an AI employee produces before relying on it or sending it to anyone; for ensuring that its use of AI employees complies with its own internal rules, confidentiality obligations, employment-law consultation duties, and the contractual and legal restrictions that apply to its data; and for informing its staff and clients as the AI Employees Notice describes. Sections 4 (Acceptable use) and 13 (Indemnification) apply to everything an AI employee does on the Customer's instruction as if the Customer had done it.
  • A human manager, and approval of consequential actions. Every AI employee must report, directly or through AI employees that lead it, to a natural person designated by the Customer who holds the "Manage AI employees" permission: the first person above it in its reporting line (an AI team lead or AI Executive on that line is passed over), who answers for every request the AI employee files. On its own virtual computer an AI employee installs software, connects software it installed there to its work sessions as a local tool server, runs commands, and edits and deletes files without anyone's approval; its calls to such software are not filed as requests. Acts that reach outside that computer need prior approval of the specific act, or are blocked. Publishing or deploying anything other than a website preview, sending e-mail outside the workspace, making a payment, deleting anything that is not on its own computer, such as a record in the Service, and raising its spending cap need that person's prior approval. Pushing code and opening a pull request, and having a pull request merged, need prior approval by that person, by a standing permission for that repository given by a person who holds the "Manage AI employees" permission and is above the AI employee, or by an AI employee in the role "AI Executive" above the AI employee that filed the request, if that AI Executive's setting described below is on. If that person does not hold the permission, the request is refused. Where the Customer leaves on the setting "Decides its team's requests" of an AI employee in the role "team lead" or "AI Executive" — on that AI employee's own page; on by default for an AI Executive and off for a team lead — that AI employee decides, under the rules and limits of the AI Employees Notice, certain requests of the AI employees below it that concern only AI employees' work on internal projects without a client, including, for an AI Executive, pushing code and having a pull request merged; an AI Executive may also decide a request to add a website to the websites an AI employee below it may reach: a website on the list of common software sources that Vezoft keeps or — since version 1.45 of the AI Employees Notice, from the day Vezoft switches it on — a website that passes TimerOS's safety check, for an AI employee of its own line, where the request names an open task of that AI employee; a website so approved goes on a list kept for the workspace and then opens for another AI employee of that line that asks for it, naming an open task of its own, where that AI Executive could decide the request itself, as an approval recorded under that AI Executive, and a person who holds the "Manage AI employees" permission can remove it from that list, as that Notice describes; a request that no such AI employee may decide goes to that person. No AI employee ever decides its own request, a request filed by or about a person, a payment, e-mail outside the workspace, publishing (a merge may still deploy through the Customer's own pipeline, see below), deleting records in the Service, a spending cap, anything about a public repository, a push that changes a repository's CI or test configuration, or a website off Vezoft's list of software sources, unless it passes TimerOS's safety check, the AI employee that asked is on the deciding AI Executive's own line and the request names an open task of that AI employee, as the AI Employees Notice describes. An AI employee's decision is carried out under the permissions of the person who answers for the request, as if that person had approved it; that person answers for it on the Customer's behalf, sees every such decision in the Service and in a daily digest, can undo it within 7 days of the decision and can take over any request an AI employee holds, and a request that no AI employee has decided within 24 of the workspace's working hours from its filing goes to that person. A person's refusal returns that exact request — the same code, merge or plan, filed again by any AI employee — to a person for good; an undo, or a refusal where that person so chooses, also returns the repository, project or AI team concerned — and the plans, work proposals and new-project requests of the AI employee that asked, and for a new project those of every AI employee — to a person for 7 days, unless a person above the AI employee who holds the "Manage AI employees" permission, or the person who chose it, ends it sooner. The deciding AI employee is instructed to pass up to a person anything that looks off. A pull request is merged only through the Service's merge request, approved as above and decided on the code host's own record of the pull request, and never by the AI employee itself: through the code host's own auto-merge where the repository's branch rules apply; otherwise, on an AI employee's decision, only when the repository's branch rules require checks, each check they require succeeded on that exact commit (a skipped check does not count), every other check on it passed and the deciding AI employee states that it is confident; in any other case a person decides. The AI employee does not wait for the answer: it files the request and carries on with other work, and an approved act is carried out at most once, exactly as it was filed, within 7 days of the decision. The request is filed by the software on the AI employee's own computer and is not checked by the Service's servers; what keeps such an act from happening without approval is described in the AI Employees Notice — among other things, the computer holds no credentials for payments, sending e-mail or deploying, and a firewall outside it limits its outbound network, though the AI employee does not ask before sending data to a website that firewall admits, such as by submitting a web form. As the deployer of its AI employees, the Customer chooses the persons who answer for them and is responsible for the branch rules and checks of its repositories, for knowing that a merge may publish code or deploy live, and for setting "A person always decides" on any project or repository whose requests it wants a person to decide; the Customer must not put the TimerOS GitHub App on any list that bypasses its branch rules. An AI employee can never be the manager of any person, an approver of a request filed by or about a person, an owner of goals over people, or a participant in any performance, activity-classification or monitoring surface concerning a person, and it cannot log or edit working time; the Service enforces these limits server-side and the Customer must not attempt to circumvent them.
  • Fees. Each AI employee is billed at the monthly rate for its tier published at https://timeros.ai/pricing, which includes a stated number of active hours per billing period; active hours beyond that are billed at the tier's published overage rate per hour, up to a monthly spending cap set by the Customer, at which the AI employee stops working until the cap is raised or the next period begins. AI employees are billed under a separate subscription from the Customer's seat subscription — in advance for the monthly rate and in arrears for overage; hiring, tier changes and removal take effect for billing at the next billing period, without proration. Prices exclude VAT and other taxes. AI-employee fees are a software-service charge, not wages; the Service records them in the Customer's Expenses in a dedicated category, and those records are system-generated and cannot be edited by the Customer. Section 3 applies to payment, renewal, refunds and taxes save as varied here. On a downgrade to a plan that does not include AI employees, every AI employee is suspended and the AI-employees subscription ends at the end of the current period.
  • Sub-processors. The model-routing gateway, the AI model providers it routes to, and the hosting provider that Vezoft engages for AI employees are Vezoft sub-processors, listed on the Sub-processors page together with their locations and the transfer safeguards. Unlike the assistants in Section 2A, they are engaged by Vezoft, not by the user. The Customer acknowledges that model inference for the bundled tiers takes place outside the European Union — no EU-region routing is available to Vezoft for them — that a request is routed to one of several model providers from an allow-list Vezoft pins, and that the gateway, the providers, the countries of processing and the transfer safeguards are as described on that page, which may change under the sub-processor notice procedure there. Personal data an AI employee processes is Customer Personal Data under Section 5 and the DPA.
  • No warranty of output. AI employees are driven by third-party models that Vezoft does not control. Vezoft gives no warranty that an AI employee's output, actions or work product will be accurate, complete, fit for any purpose, free of third-party rights, or produced within any time; Section 11 applies. The Customer's remedy for an unsatisfactory AI employee is to pause, correct or remove it.
  • Suspension. In addition to Section 10, Vezoft may pause or suspend any AI employee, or the feature as a whole, at any time where it reasonably considers this necessary for the security or integrity of the Service, to prevent abuse or harm to third parties, to enforce these Terms or the Acceptable Use Policy, to stay within the Customer's spending cap, or to comply with law.
  • Enterprise bring-your-own-key. Where an Enterprise Customer runs AI employees on its own model-provider account, the Customer contracts with and pays that provider directly, Vezoft charges the platform fee and the virtual computer, and the model provider is the Customer's processor, not Vezoft's sub-processor, for that Customer.

3. Subscriptions, payment, and trial

  • Pricing for each plan is published at https://timeros.ai/pricing. We may change pricing on 30 days' notice for the next billing cycle.
  • Free trial: 14 days. We collect a payment method during trial via Stripe; we do not charge until the trial ends. Cancel at any time during the trial to avoid charge.
  • Subscriptions renew automatically each billing cycle until cancelled. Cancellation takes effect at the end of the current paid period.
  • Refunds: 14-day money-back guarantee on first purchase. After that, refunds are at our discretion and as required by mandatory consumer-protection law in your country.
  • Taxes: prices exclude VAT, GST, or sales tax unless stated. We charge applicable taxes based on your billing address.
  • Payment processing: by Stripe. We do not store full card numbers. Stripe is our sub-processor and acts as a separate controller for fraud prevention.
  • AI employees: billed under a separate subscription at the per-tier monthly rate, with the included active hours, overage rate and spending cap described in Section 2B; changes take effect at the next billing period.

4. Acceptable use

You agree to comply with the Acceptable Use Policy, which is incorporated by reference. In short: no illegal use, no harming the Service or others, no use of TimerOS to violate workers' or third parties' rights.

5. Data processing — controller and processor relationship

For personal data of your employees, contractors, clients, and other individuals whose data you upload to or generate in the Service ("Customer Personal Data"), you are the data controller and TimerOS is the data processor. Our processing of Customer Personal Data is governed by the Data Processing Agreement ("DPA"), which is incorporated by reference and which you accept by accepting these Terms.

For your own account information (your name, email, billing info), TimerOS is the controller. See the Privacy Policy.

6. Customer responsibilities for employee monitoring (this clause is load-bearing — read carefully)

You acknowledge that:

(a) The Service tracks time, work activity, and (where you enable it) activity classifications generated by a local AI classifier on the worker's device.

(b) Workplace monitoring is regulated under the laws of the workers' country of residence and place of work. Examples include but are not limited to: GDPR Art. 88; national employment privacy laws in Germany (BDSG §26), France (Code du travail L1121-1 et seq.), Italy (Statuto dei Lavoratori Art. 4), Spain (Estatuto de los Trabajadores Art. 20.3), the Netherlands (Wet bescherming persoonsgegevens / works-council consent under WOR), Austria, Belgium and Poland; in the United States, statutes including but not limited to N.Y. Civil Rights Law §52-c, Conn. Gen. Stat. §31-48d, Del. Code §19-705, and California Labor Code §435 and §638(g).

(c) You — not TimerOS — are responsible for:

  • providing written notice (the Employee Monitoring Notice) to your workforce, in their language, before activating activity detection in the desktop application;
  • obtaining consent or fulfilling any other lawful basis required in their jurisdiction;
  • consulting and obtaining the agreement of works councils, employee representatives, or trade unions where required by national law;
  • maintaining records of consent and consultation as required by national law;
  • acknowledging in the TimerOS admin console, at the point where activity detection is enabled, that the Employee Monitoring Notice has been delivered to the affected workforce. This acknowledgement is recorded on your tenant with the acting admin's user ID, timestamp, IP address, and the notice version, and is binding on the Customer organisation.

A template Employee Monitoring Notice is provided at policies-customer-facing/employee-monitoring-notice-template.md for your convenience. The template is not a substitute for local legal advice.

Solo customers. Where you are an individual customer on the Freelancer plan and the only person whose activity the Service tracks, you have no workforce to notify, and the obligations in (c) do not apply to your own use. Your own transparency is addressed directly in-product: during onboarding, before activity detection is first enabled, the Service presents you with an activity-tracking disclosure describing what is tracked and how the on-device classifier works, and records your acknowledgement. If you later add any worker other than yourself to your workspace, the obligations in (c) apply in full from that moment, before activity detection is enabled for that worker.

(d) You agree not to use the activity classification output of the Service as the sole basis for any decision affecting a worker's employment, pay, discipline, demotion, promotion, or termination. The classification is provided for self-observation, resource allocation, and identification of broad team trends. Any employment-related decision must be made by a human, based on a wider body of evidence and following the procedures required by your local labour law.

If you fail to comply with (a)–(d), you indemnify TimerOS for any resulting claims, fines, or losses, including under Section 13.

6A. Manual time-entry editing

The Service permits Customer's administrators and authorised users to create, edit, correct, and delete time entries after the fact. Customer is solely responsible for ensuring that such edits are made in accordance with applicable wage-and-hour, working-time, payroll, and recordkeeping law in each jurisdiction where its workers perform work (for example, the U.S. Fair Labor Standards Act and state equivalents, the EU Working Time Directive and national transpositions, and any applicable collective-bargaining agreements). Customer must not use the edit functionality to falsify records of hours worked, to suppress overtime, or otherwise to defeat statutory protections for workers.

TimerOS records every create, edit, and delete action against a time entry in an append-only audit log — no TimerOS interface (UI or API) can edit or delete an entry; this is enforced at the application and database layer (there is no PATCH or DELETE route for audit-log records, and the ORM layer rejects in-place updates and deletes via pre-hooks). It is not storage-level immutable (a privileged operator with direct database access could in principle alter the raw collection), so TimerOS does not claim cryptographic or WORM immutability. The log captures the before/after values, the acting user, the timestamp (UTC), the source IP, and the reason field (where supplied). The audit log is retained for the life of the tenant plus the retention period set out in the DPA and is accessible to Customer's administrators through the admin console. TimerOS does not modify Customer time entries and provides no warranty that Customer's editing practices comply with local law — that responsibility, and any liability for non-compliance, rests with Customer.

7. Service availability

We operate the Service to a non-binding monthly uptime target of 99.5% for paid plans, as set out in the SLA. For the launch period this is an operational target only: it is not a contractual guarantee and carries no service credits. Binding uptime commitments and a service-credit remedy are reserved for a future Enterprise SLA. Free-trial accounts are provided "as is" without any availability commitment.

8. Intellectual property

We grant you a worldwide, non-exclusive, non-transferable, revocable licence to use the Service during your subscription. All right, title, and interest in the Service and in our trademarks remain with TimerOS.

You retain all rights in Customer Personal Data and in your tenant content. You grant us a worldwide, non-exclusive licence to host, process, transmit, copy, and display Customer Personal Data and tenant content only to the extent required to provide and improve the Service, and only in accordance with the DPA. We do not use Customer Personal Data to train any central or cloud AI models we hold. The on-device activity classifier is a scored rule engine combined with a small on-device statistical model that ships with fixed seed weights and may adapt over time by retraining locally, on the worker's own device, from that user's bounded manual corrections. This on-device learning stays on the device: no classifier training data, corrections, or telemetry are sent to TimerOS. For a given installed version together with the device's local corrections state, the classifier's behaviour is deterministic and reproducible. We maintain a model card and an EU AI Act applicability assessment for the classifier, available to competent authorities on request.

Feedback you give us about the Service is non-confidential, and we may use it without restriction.

9. Confidentiality

Each party will protect the other's Confidential Information with at least the same care it uses for its own confidential information (and no less than reasonable care). "Confidential Information" excludes information that is or becomes public through no fault of the receiver, was already known, is independently developed, or is disclosed lawfully by a third party.

10. Suspension and termination

We may suspend or terminate your access if you materially breach these Terms (including non-payment) and fail to cure within 14 days of written notice, or immediately for any breach of Sections 4, 6(d), or for legal compliance reasons. You may terminate at any time by cancelling your subscription; access continues to end of paid period. On termination we delete Customer Personal Data per Section 12 of the DPA.

11. Warranty disclaimer

To the maximum extent permitted by law, the Service is provided "as is" and "as available." We disclaim all implied warranties including merchantability, fitness for a particular purpose, and non-infringement. We do not warrant that the Service will be uninterrupted, error-free, or that the AI classifier outputs, or the output of an AI employee (Section 2B), will be accurate. Mandatory consumer-protection rights (where you qualify as a consumer) are not affected.

12. Limitation of liability

To the maximum extent permitted by law:

  • Neither party is liable for indirect, incidental, consequential, special, punitive, or exemplary damages, or for loss of profits, revenue, goodwill, anticipated savings, business opportunity, or data — even if advised of the possibility.
  • Our total aggregate liability for any claim arising out of or related to these Terms or the Service is capped at the amounts you paid us in the 12 months preceding the event giving rise to the claim.
  • Nothing in this Section limits liability for fraud, gross negligence, wilful misconduct, death or personal injury caused by negligence, or any liability that cannot be limited under applicable law (including liabilities under GDPR Art. 82 to data subjects which run directly under the regulation).

13. Indemnification

You will defend, indemnify, and hold harmless TimerOS from and against any third-party claims, costs, damages, losses, liabilities, and expenses (including reasonable legal fees) arising out of or related to:

(i) your breach of Section 6 (Customer monitoring responsibilities) or the DPA; (ii) your tenant content infringing the rights of any third party; (iii) your use of the Service in violation of any law; (iv) any employee/worker claim against you arising from your monitoring activities, where your handling of notice, consent, or consultation was deficient.

14. Governing law and jurisdiction

These Terms are governed by the laws of Bulgaria, excluding its conflict-of-laws rules and the UN Convention on Contracts for the International Sale of Goods. Each party submits to the exclusive jurisdiction of the courts of Kardzhali, Bulgaria. EU consumers retain the mandatory right to bring claims in the courts of their country of residence and the protection of the mandatory consumer-protection law of that country.

15. Changes to these Terms

We may update these Terms; changes take effect on the date stated when posted. We will give the account owner at least 30 days' notice of material changes (changes that reduce your rights, increase your obligations, or change pricing) by email and in-app. If you do not accept the change you may terminate before it takes effect.

16. Miscellaneous

  • Entire agreement. These Terms, the Privacy Policy, the DPA, the AUP, the SLA, and any order form together form the entire agreement between us, and supersede prior agreements on the same subject matter. In case of conflict the order is: order form > DPA > Terms > AUP > Privacy Policy > SLA.
  • Assignment. You may not assign these Terms without our consent. We may assign in connection with a merger, acquisition, or sale of assets.
  • Severability. If any provision is unenforceable, the remainder remains in force.
  • No waiver. Failure to enforce is not waiver.
  • Force majeure. Neither party is liable for delays caused by events beyond reasonable control.
  • Notices. To us (Vezoft EOOD, operating TimerOS): support@timeros.ai, with a postal copy to our registered office (see Imprint for the full address). To you: the email address on your account.
  • Export controls. You confirm you are not on a US, EU, or UK sanctions list and are not located in a comprehensively sanctioned country.

Contact: Vezoft EOOD, Bulgaria — see the Imprint for our full registered address and Unified Identification Code (UIC). Email support@timeros.ai.