A policy gate on every action
Before an AI worker sends a message, books, spends or touches data, Ordibl checks it. If the check cannot be completed, the answer is no.
Canada · United States
Ordibl's AI workers answer customers, follow up on leads and handle routine operations. Every action is checked against the law, your industry and your own policies first. Risky steps wait for a person. Everything is on the record.
Pilot program open in Canada and the United States. The chat on this page is a scripted preview, not a live phone line.
The problem
Your customers call after hours. Your team is stretched. And most AI tools cannot tell you what they will say, or prove what they did. For a clinic, a lender or a housing office, that is not a risk worth taking.
After-hours calls and unanswered messages leave work, and revenue, on the table.
If you cannot see the rule behind a reply, you cannot put that AI on the phone or in the chat.
Without an audit trail, a complaint or a regulator leaves you reconstructing the day from memory.
How it works
Choose your region and industry, then add your own policies.
Answering, booking and following up, inside the rules you set.
The policy gate allows it, blocks it or asks for approval. When rules disagree, the strictest one wins. If a check cannot run, the answer is no.
Approve risky steps, take over any conversation, and review the full record.
AI workers
Answers common questions from your knowledge base and hands off when it is unsure.
Continues a conversation you already started. It does not invent a campaign, and it does not send mail you did not approve.
Handles routine internal steps that still have to pass the policy gate and stay on the record.
Channels
Web chat is in the product. SMS for Canada and the United States, and phone connectivity, are in the pilot build and stay switched off until a pilot turns them on. This page does not publish a number to call.
WhatsApp is not offered on this site. The preview above never places a call or sends a text.
Governance
Built-in safety rules no one can switch off. Regional layers for GDPR, PIPEDA and US rules. Your industry and your policies on top. When rules disagree, the strictest wins.
Before an AI worker sends a message, books, spends or touches data, Ordibl checks it. If the check cannot be completed, the answer is no.
High-risk steps wait for a named person to approve, reject or edit. The person who asks does not approve their own request.
Every decision and action is recorded in a history the app cannot edit or delete, with the rule that applied and a plain-language reason.
Each organization can have a hard AI spending limit, reserved before a call to a model, so usage stays inside that limit.
Each organization is isolated at the database level. One customer cannot read another's data.
Low confidence, or a language the worker should not handle alone, hands the conversation to your team. Customers are told they are talking to an AI.
Languages
Eight languages: English, Spanish, French, Arabic, Yoruba, Hausa, Igbo and Pidgin. These lines are samples on this page. A language is served to customers when it has been reviewed. If Ordibl is not sure, it hands the conversation to your team.
On the record
Text recognition for these eight languages is in the product. We do not claim finished voice quality for every language, and this site does not place a call.
Industries
No customer logos. These are the workflows the pilot is meant to fit.
Hours and bookings, with a person confirming anything clinical. Do not put real health information in the demo.
Approvals and spending caps before the AI moves money or makes a promise.
An audit trail of what was said about a unit, a repair or a viewing.
Reservations and guest questions, with a person when a stay, a complaint or a charge needs a decision.
Shipment updates and routine exceptions, with a person when a delivery is disputed or a promise would change.
Features
English, Spanish, French, Arabic, Yoruba, Hausa, Igbo and Pidgin, with a handoff when the language is unclear.
Answers come from what you have written. If it does not know, it says so and brings in your team.
Bookings and anything you mark as sensitive wait for a person on your team.
When a customer asks for a person, your team can take the conversation with its history.
Owners, admins and staff each see only what their role allows.
Export, correction and deletion requests, legal hold and retention rules are part of the product, not a slide.
Pricing
Plans and prices are not published yet. There is no fee on this page. When pricing is published, you will be able to pay in your local currency.
Company
Ordibl is the flagship product of AIVIRA, headquartered in Regina, Canada.
FAQ
Each organization is isolated in the database. A data region can be assigned for a pilot. This site does not promise a hosting country. The scripted chat above stays in your browser.
Only inside the rules you set. Risky actions wait for a named person. If the policy check cannot run, the action does not happen.
It says so, or it hands the conversation to your team. Customers are told they are talking to an AI and can ask for a person.
No. We do not hold SOC 2, ISO 27001 or HIPAA certification. See the box in the governance section, and the trust page.
Eight: English, Spanish, French, Arabic, Yoruba, Hausa, Igbo and Pidgin. Greetings on this page are samples. Voice quality is not a finished claim.
We have not published a price. Paying in your local currency is the intention when pricing exists. Book a demo and we will agree the pilot with you directly.
Pilot
Tell us about your organization. We will show you the rules, the approvals and the audit trail. The pilot suits teams in Canada and the United States. Length and price are agreed with you. There is no self-serve sign-up.