Ask your cloud anything
Answers in plain language, grounded only in your collected evidence. Every number in an answer must appear in that evidence, or VIREOPS shows the evidence instead.
The questions your team already asks
These are the suggested questions on the Ask VIREOPS screen of our demo estate. Each answer is built from the same facts the dashboards use.
Read from collected billed cost, by service, compartment and region.
Read from framework results and the failing checks behind them.
Answered with a network-access check where code, not the model, decides whether a rule allows a port.
Read from the same deterministic recommendations as Optimize.
Read from the inventory, scoped to what you are allowed to see.
Read from the latest collected state of each instance.
Your model, your region, your key
VIREOPS does not ship a model or route your questions through NotionAce. You choose the provider, and an administrator switches it on.
- OCI Generative AI in any subscribed region.
- Your own Anthropic API key, stored encrypted.
- Off until an admin enables it, with a daily token ceiling.
- Residency shown plainly: the screen says where questions and facts are processed.
- No silent fallback to another provider.
Built so the model cannot make things up
The model explains and correlates evidence. It never computes a score or a total, and it cannot reach anything the person asking cannot see.
Numbers verified
Every figure in an answer must appear in the evidence it was given. If one does not, VIREOPS shows only the evidence.
Read-only by design
Answers come from read-only tools that run with the asker's own permissions. No SQL, no shell, no writes to the cloud.
Code decides, not the model
Scores, totals and network reachability are computed by code. The model never computes a score or a total.
Complete without it
The product is complete with AI switched off, and the build checks that it stays that way.
Real follow-up questions, every fact read again
Ask a follow-up and the conversation continues, with the last four turns as context. The facts are re-read for every answer rather than trusted from earlier turns.
- History per user, so each person sees their own conversations.
- Every question is audited.
- Licensed as the AI module, so it can be left out entirely.
Ask VIREOPS at a glance
| Capability | Detail | Status |
|---|---|---|
| Providers | OCI Generative AI in any subscribed region; your own Anthropic API key | Live |
| Enablement | Off until an administrator enables it; daily token ceiling | Live |
| Grounding | Read-only tools in the asker's permissions; every number checked against the evidence | Live |
| Network access answers | Code decides whether a rule allows a port | Live |
| Conversations | Last 4 turns as context; facts re-read for each answer; per-user history; every question audited | Live |
| Semantic layer | Ontology, query DSL and metric registry behind Ask VIREOPS | Not yet |
Questions about Ask VIREOPS
Do our questions go to NotionAce?
No. Questions go from your own VIREOPS install to the provider you chose. NotionAce never receives users, AI questions or AI keys.
Can we keep AI processing inside our region?
Yes. Choose OCI Generative AI in a subscribed region that meets your residency needs. The AI settings screen shows where questions and facts are processed.
Can the AI change anything in our cloud?
No. Its tools are read-only and run with the asker's own permissions. It has no SQL, shell or cloud-write tools.
What happens if the model invents a number?
Every figure in an answer must appear in the evidence. If one does not, VIREOPS does not show the answer; it shows only the evidence.
Do we need AI to use VIREOPS?
No. AI is an optional module, off by default. Every dashboard, score, recommendation and report works without it.
Let's map your cloud estate together.
See VIREOPS on a live OCI estate: scored, priced and evidenced in one walkthrough. NotionAce deploys it in your tenancy within days.