ANVE: the private AI office, on the record. See what that means
ANVEPrivate AI glossary · updated 2026-10-07
Short answer. A custom GPT is a version of ChatGPT that someone configures for a specific purpose by writing instructions, attaching documents as knowledge and, optionally, connecting it to outside services through actions. OpenAI’s documentation describes custom GPTs in those terms. It runs inside ChatGPT, so the files and conversations sit in OpenAI’s service, under the workspace’s settings. As of October 2026, OpenAI’s guidance for enterprise workspaces says it is transitioning custom GPTs to plugins and that custom GPTs will stop working at retirement; the page gives no date and says timelines for other plans may differ. An office that wants an assistant tied to its own documents, on its own hardware, can run a private GPT instead.
| Part | What it is | Office example |
|---|---|---|
| Instructions | Written guidance on role, tone and rules | You are the front desk for a dental clinic and quote fees only from the attached list |
| Knowledge files | Documents attached for the GPT to draw on | Fee schedule, consent forms, FAQ |
| Actions | Calls to third-party services, defined by a schema | Check appointment availability in the booking system |
| Sharing | Who may open it: only the builder, named people or the workspace | The clinic’s staff only |
A custom GPT is a feature of a hosted service. Its instructions, attached files and chats are handled by the provider, within the controls the workspace administrator sets. OpenAI’s documentation for enterprise workspaces describes controls over who can create and share GPTs, over connected apps and custom actions, and an approved-domains list that restricts which outside services a GPT’s actions may call. If no domains are allowed, custom GPT actions cannot run.
Before an office puts client material into one, it should read the provider’s current data-handling terms for its plan and decide whether that material may leave the building. Our questions pages on ChatGPT and company data cover storage and training.
OpenAI’s guidance for enterprise administrators says it is transitioning custom GPTs to plugins in three milestones: an admin notice, a migration period, then retirement, when custom GPTs stop running and leave the GPT directory. Once a GPT is migrated, the original becomes read-only. Instructions and reference files can carry over to a plugin as a skill, but custom actions do not transfer and must be replaced, for example with a supported app or a custom MCP server. The page gives no retirement date, and says timelines for other plans may differ.
The lesson is general. An assistant built on a vendor’s platform lives on the vendor’s roadmap. Keep the instructions and the source documents in your own files, so that you can rebuild elsewhere.
A private GPT does the same job on infrastructure the office controls: instructions, the office’s own documents and an open model, running on a machine in the building. When comparing options, check:
ANVE’s private GPTs are assistants built on the office’s own material, and ANVE answers from documents with the source passage shown. ANVE’s security page (/security) says every order includes a perpetual licence for the software delivered with the unit, and that the box keeps working if ANVEAI shuts down. ANVE is pre-release.
No sealed ANVE unit exists; every image of the appliance on this site is a labelled render. The software runs today on a development machine. No customers yet, no certifications, no delivery date, and the final price is not announced.
Reservations are ₹999 in India or US$12 elsewhere, refundable in full until you confirm an order in writing (terms). Offices that cannot use cloud AI can apply to be one of ten design partners.