VaultOS Assistant

One place to ask.
Grounded, cited, and gated — for people and agents.

We gave our whole company one URL to ask. Yours runs on your own domain — literally ask.yourcompany.com, with its own certificate, not a link on ours — grounded on a governed vault of everything your organization knows. Every answer cites the exact source note, so it can be checked rather than trusted.

Book a free walkthrough

Try the live demo

See how it works

Where your knowledge lives today

Your knowledge doesn't live in one place. It's in SaaS tools, a couple of clouds, on-prem systems — and mostly in people's heads. None of your AI can see across all of it. Neither can your new hires.

SaaS tools and drives

Docs, decisions, and runbooks spread across Notion, drives, and email. Each tool holds a slice; none holds the picture.

A couple of clouds

Context split across cloud platforms — and every AI project re-solves the same context problem from scratch, because none of your tools can see across them.

On-prem systems

Records and processes in systems your AI never meets — invisible to every assistant you're paying for.

People's heads

"When she left, fifteen years of process knowledge walked out with her." Nobody disagrees with it. Everybody has a version of it.

So we built the bridge

Three layers turn scattered knowledge into one brain your people and your agents can actually ask.

Layer 01

One governed vault

Everything converges into one governed vault — plain, open-format files you own, linked into a graph. Sync to every device is end-to-end encrypted; the sync server only ever holds ciphertext.

Layer 02

Catalogued in Google Cloud

The vault's manifest lands in BigQuery, and Knowledge Catalog registers every note automatically — governed, discoverable metadata for the whole graph.

Layer 03

Grounded in Gemini Enterprise

The notes index into a dedicated search data store that Gemini Enterprise grounds on. Ask in plain language — every answer cites the exact notes it came from.

VaultOS assistant answering a question with inline citations and a grounded-sources list — product illustration

What it feels like

Ask in plain language. Get the source.

No prompt engineering, no hunting through drives. Your team asks the way they'd ask a colleague — and every answer comes back grounded in your own notes, with the exact sources cited inline. If there's no source, there's no claim.

It runs on your domain, not ours

Most "your own assistant" products hand you a page on the vendor's website. Yours answers at ask.yourcompany.com — your domain, your DNS record, your certificate, your name in the address bar. Two client-owned domains are serving in production today.

What that changes in practice:

Your people recognise it. No one has to trust a link on someone else's domain to ask an internal question.
The exit is cleaner. Your vault already exports with a single git clone. Now the address your team knows it by belongs to you too — the DNS record is in your zone, and you can point it elsewhere without asking us.
Setup is one record. You add a CNAME. We issue and renew the certificate automatically, and the gate sits in front exactly as it does on ours.

Gated by default. Cited by design.

The assistant is gated, and it runs on your own domain — and the sync layer underneath it is end-to-end encrypted.

Gated before it loads

Every request is checked at the edge before the assistant loads. It lives behind that gate on your own domain — only signed-in, authorized people reach it.

Scoped views fail closed

Scoping is enforced server-side: each view grounds only on the areas you allow. A partner literally cannot retrieve outside their slice — an out-of-scope query finds no sources.

Every answer cited

Every answer cites the exact source note it came from. No citation, no claim.

A dedicated, removable store

Grounded-assistant indexing is opt-in per tier. Your notes index into a dedicated per-tenant data store — scoped, access-gated, and removable whenever you choose.

The same brain serves two audiences

Same vault. Same governance. Human or agent, nobody works from memory anymore.

Your people

Plain-language asks, answers with receipts

Ask in plain language and get decisions back with receipts — every answer links the exact note it came from. Save new decisions back into the vault, so the graph compounds.

Your agents

Grounded briefs, policy checks, write-back

Agents pull grounded briefs, run policy checks before acting, and write new knowledge back into the vault — under the same governance as everyone else.

Run in production first

1,038 documents across six businesses run through this today — our own operation is the reference install, not a demo environment.

1,038

documents indexed for Gemini Enterprise grounding

0

import failures on the last full sync

100 %

of answers cited — sources link to the exact note

6

businesses operated on one graph

Give your AI a memory.

A free 30-minute walkthrough of the live system — the graph, the sync, and the assistant that cites three years of decisions back to you. It's the live install, not a demo environment.