Frequently asked questions

OpenExpertsAI is a private AI workspace for organisations with sensitive data. Specialist assistants work from your own documents and reference library, and cite the source in their answers. The AI models run on the server that hosts your workspace, so documents and chats are not sent to a public AI service unless you allow it. We are based in Ho Chi Minh City, Vietnam.

Any organisation that needs AI but cannot let its data out. Examples are law firms, health clinics, laboratories and research teams, as well as finance and HR teams. See how it works for law firms, clinics, labs and research teams.

Your team can ask questions about uploaded contracts, medical records or reports, and a source panel shows the passage behind each answer. Recordings of meetings, consultations or interviews are transcribed on the server and drafted into notes for a person to edit and confirm. Before a document is shared, the redaction assistant finds names, ID numbers and other personal details for a person to review, then exports a redacted copy. See all features.

Open-weight models, for example from the Qwen family, that run on the server itself. Your documents, chats and recordings are processed on that server. If your organisation chooses, it can also allow some accounts to use a stronger outside AI.

Documents, chats and recordings are processed on the server that hosts your workspace; two switches can send something out, and which ones are on is agreed with you during setup. Web search sends search queries made from a user's question to a web search service: documents are not sent, but a query can contain words from the question, such as a name. Outside AI is off unless your organisation chooses it; when on, it sends the conversation, including any document passages in it, to an outside provider after common personal details are automatically replaced with placeholders, but the masking is not perfect and does not look for unpublished figures or results. Separately, if your workspace is reached through Cloudflare, Cloudflare's network decrypts and re-encrypts the traffic on its way to the server, possibly outside Vietnam. Read more about security and data handling.

No. OpenExpertsAI does not use client data to train AI models. The models on the server do not learn from your documents or chats. Outside AI is a separate, optional service; see the question on outside AI.

Yes. Each organisation gets its own workspace, with its own application instance, database, document libraries and user accounts; clients' documents are not mixed. During a pilot, your workspace runs on a server OpenExpertsAI operates, which can host other organisations' pilots too, each in its own separate workspace.

The people your organisation gives accounts to: each person signs in with their own account, and your administrator decides who can use which assistants and features. During the pilot and in a workspace we host, OpenExpertsAI's administrators can also access the server to maintain and support it; on a server in your own office, your organisation controls who can access it. Hosted workspaces are reached through a tunnel service, and if yours is reached through Cloudflare, Cloudflare's network decrypts and re-encrypts the traffic on its way to the server, possibly outside Vietnam. If your organisation turns on outside AI, the outside provider also receives the conversation, including any document passages in it, after personal details are masked; see the question on outside AI.

Yes, as an option: it is off unless your organisation chooses it, and only the accounts your administrator allows can use it. What goes out is the conversation, including any document passages in it, after personal details such as names, ID numbers, phone numbers, bank account numbers and addresses are automatically replaced with placeholders; the real values are put back in the answer on the server. The masking looks for personal details, not unpublished figures or results, and it catches common forms but is not perfect, so we agree staff rules with you for what may be sent.

We use them to contact you about the demo and a possible pilot, and we do not add you to a mailing list. They are stored on our own server in Vietnam and are not sold or shared with other companies. We delete them 24 months after our last contact with you, or sooner if you ask. Read our privacy policy.

Yes. After the pilot, you choose between a workspace we host and operate for you and a server in your own office. With a server in your office, your organisation controls who can access it. During the pilot itself, the workspace runs on a server we operate.

First, we have a short call about the work that takes your team the most time and the data that needs protecting. Next, we set up a private workspace for your organisation on a server we operate, load your documents and add accounts for your team. Then your team uses it on real work for one month while we adjust the assistants to how your organisation works. Finally, you choose where it runs: in a workspace we host or on a server in your own office.

We do not publish prices. The cost depends on the number of users, the hardware and whether we host the workspace for you. We give you a quote after the first call.

We delete your workspace and its data, and confirm the deletion to you in writing. We also agree backup, retention and deletion arrangements with you in writing.

Fill in the demo request form and we will get in touch to find a time. The demo takes 30 minutes. Bring a document your team works with, and we will show how the workspace handles it.

Each answer cites its source, and the source panel shows the passage behind it. Answers can still be wrong, so a responsible person should check them before they are used. The assistants do not diagnose, give legal advice on their own or replace a professional's sign-off.

The interface is in English and Vietnamese. Law firm workspaces can include a library of Vietnamese legislation. If your documents are in another language, bring one to the demo and see how the workspace handles it.

Yes. It works in a web browser, including on a phone. Each person signs in with their own account.

Yes. In team channels, colleagues discuss a matter, case or study, and bring an assistant into the thread by typing @ and its name.