AI and your data

The Services include VO, an AI assistant. When you use it, the content needed to answer your request — your question, and the records from your own Workspace that your role is permitted to read — is sent to a language model provider over an API, and the answer is sent back.

Three things are worth being precise about:

• It is sent to answer you and for nothing else. We do not permit our providers to use it for model training, and our agreements with them reflect that. VO does not learn from your data and is not trained on it.
• VO cannot exceed your own permissions. It reads what your role can read. If a record is invisible to you, VO cannot retrieve it, summarise it, or act on it.
• Speech is transient. Where you use voice, audio is processed to produce or interpret speech and is not retained by us afterwards.

The providers are listed on the subprocessors page. If you would prefer no AI processing at all, the entry VO tier can be left unused, and an Enterprise agreement can disable VO for your Workspace — write to contact@ofivio.com.

The Services are built on third-party infrastructure and call third-party APIs. This is not incidental — it is how VO, voice, file storage and authentication work, so you should know it before you rely on them.

• Hosting, database, authentication and file storage run on Google Cloud / Firebase.
• VO's language models are provided by Anthropic, OpenAI and DeepSeek. Which one answers a given request depends on the task and on availability; we may change providers.
• VO's speech is generated by ElevenLabs.
• Sign-up and sign-in are protected by Google reCAPTCHA.

When you use one of these features, the content needed to serve that request is transmitted to the relevant provider. It is sent for the purpose of answering you and is not contributed to any provider's training data. The current list, and what each one receives, is in the Privacy Policy.

What VO is, and is not

VO can read what your role is permitted to read, and at the highest tier it can act — creating tasks, drafting documents, moving work along. It acts as you, within your permissions. It cannot see or do anything you could not see or do yourself, and an action VO takes on your instruction is your action.

VO output is not guaranteed correct. It produces drafts, summaries and suggestions from your data, and it can be wrong. Check anything that matters before you act on it. See the Disclaimer.

Usage limits

VO is metered as a daily token budget per seat, set by your VO tier and reset each day at midnight Pakistan Standard Time. It is a daily allowance, not a monthly one — a seat that exhausts its budget regains it the next day. The Services also apply rate limits to protect availability for everybody; the current limits are published on the Security page. Fair, ordinary use will not meet them.

We use the following providers to run the Services. Each receives only what it needs for its function, under contract, and none of them is permitted to use your data for its own purposes.

Which language model answers a given request depends on the task and on availability. We may add or change providers; where a change materially affects how your data is processed, we will update this page and give notice as described in the Privacy Policy.

We may also disclose data where the law requires it — a valid court order or a lawful request from an authority — or to establish or defend a legal claim. Where we are permitted to tell you, we will.