Built-in intelligence
Machine learning that points. Generative AI that helps. People who decide.
Intelligence is part of the platform, not an add-on. It works from your own records, respects each person’s access and never acts alone.
Two kinds of intelligence, one rule
Machine learning
Forecasts spend and flags unusual attendance and invoices for review. A model is trained, evaluated and approved by a second person before it is switched on, and it can be rolled back.
Generative AI
Ask the VOC assistant questions in plain language, or connect the AI assistant you already use through the open MCP standard. It sees only what the person asking is allowed to see.
Always a person in charge
Insights advise; they never act alone. Anything the assistant would change is shown as a preview first and needs approval, and every action lands in the audit trail.
What asking looks like
Examples of the kind of question the assistant answers, drawn from the person’s own permissions.
Who can cover Friday night at the depot?
Two night shifts are short. Three people hold the right credentials, are free, and stay within rest limits. Want me to draft a request for them to apply?
Yes, draft it.
Draft ready for your review. Nothing is sent until you approve it.
Four safeguards, built in
- Access. The assistant sees only what the person asking is allowed to see.
- Preview. Anything it would change is shown first.
- Approval. A person approves before a change is made.
- Audit. Every action is recorded against a person.
Questions about the AI
Does the AI make decisions on its own?
No. Forecasts and flags tell you where to look, and every one is labelled. The assistant only sees what the person asking is allowed to see, and anything that would change your records is shown as a preview and needs a person’s approval first.
Can I use the AI assistant I already use?
Yes. People can connect the assistant they already use through the open Model Context Protocol (MCP). It acts with the same permissions that person already has, and every action is recorded in the audit trail.
How are machine learning models switched on?
A model is trained and evaluated, then approved by a second person before it is switched on. It can be rolled back.
Can the assistant change my data?
It can read what the person asking may read. Anything it proposes to change is shown as a preview first and needs approval, and the change is recorded against that person.
Ready to see it with your own organisation?
Create your organisation in a few minutes. No card needed to start.