A chat assistant your customers can use — and your firm can stand behind.
In plain English: an AI chat for your customers that follows your rules and hands over to a human when it matters.
Vör is a governed AI chat widget for your website, app or portal. Every message is checked against your policies in real time, risky conversations are escalated to a person, and each interaction leaves a reviewable evidence trail — so customers get help while your firm stays in control.
Vör
Online · governed by norse3
Policy in real time
Checked before every reply
Human escalation
A person takes over when it matters
Reviewable evidence
A trail for QA and compliance
Customers want fast answers. Your firm can't hand over the controls.
An unconstrained chatbot can give unsuitable answers, leak information or handle a vulnerable customer badly — and leave you with no record of what happened. Native model guardrails aren't your policy.
Ask → relay → policy → act → evidence.
The same control and evidence path as the rest of the norse3 layer, wrapped in a chat experience your customers recognise.
Customer asks
A visitor or customer chats with Vör on your site, app or portal.
Bifröst relay
The conversation is routed through norse3 rather than straight to a model.
Aegis policy
Each message is checked against your organisation's policies in real time.
Allow · warn · redact · block · escalate
Configured actions apply before a response is delivered.
Heimdall evidence
The event, policy, action and any human decision are retained for review.
A helpful assistant, kept on a governed leash.
Vör is designed to support alignment with your policies. It is not a compliance replacement, and your firm remains accountable.
Policy in real time
Every message and response is evaluated against your configured, versioned policies before delivery.
Sensitive-data awareness
Detect and redact designated information so it isn't echoed back or stored in the clear.
Human escalation
Hand off to a named reviewer when a conversation needs a person, not a model.
Reviewable evidence
A linked, time-stamped trail of what happened — useful for QA and compliance teams.
Model-independent
Vör governs the interaction and stays independent of the underlying model or vendor.
On-brand
Match your colours, tone and greeting so Vör feels like part of your product.
Wherever your customers already talk to you.
Public website
Answer questions and guide visitors with oversight in place.
In-app assistant
Support signed-in users inside your product.
Help centre
Deflect routine queries while keeping risky ones supervised.
Customer portal
Assist account holders within a governed boundary.
A record behind every conversation.
Where configured, Vör retains a reviewable trail of what was asked, the policy applied, the action taken and any human decision.
- Interaction reference and risk classification
- Policy applied and action (allow / warn / redact / block / escalate)
- Human escalation and reviewer decision
- Reference retained — not the raw sensitive content
Drops into your product.
Embed on your site, app, help centre or customer portal.
Hosted or private cloud
Chosen against sensitivity and data residency.
Model-independent
Governs the interaction, whichever model sits behind it.
On-brand
Match your colours, tone and greeting.
Common questions.
Does Vör replace our AI model?
No. Vör is an independent control layer in front of the model. The model stays separate — Vör controls what it can see, say and do.
Can it stop the AI giving advice it shouldn't?
Yes, where configured. Responses can be held, redacted or escalated to a person before they reach the customer.
What happens with a vulnerable customer?
Vör can detect indicators and escalate the conversation to a named reviewer under your policy.
Do you store conversation content?
Retention is configurable; Vör can operate on a minimal-evidence basis, keeping a reference rather than raw content.
Put a governed assistant in front of your customers.
See how Vör applies your policies, escalates to people and captures evidence — in a 30-minute walkthrough with our team.