Trust and security

AI that talks to your candidates has to be governed, isolated and accountable.

Q keeps people in charge of what it learns, keeps each institution's data separate, and is being independently assessed for AI risk before production.

Request the security overview

PDF, sent to your work email.

01

Governance

Q explains its reasoning, flags inconsistencies, routes low-confidence questions to advisors, and learns only what a person approves. Ranking changes need a director's sign-off. Every answer and handoff is on the audit trail.

02

Data isolation

A multi-tenant architecture in which each institution's candidates, content and conversations are isolated. No institution can see another's data.

03

AI risk assessment

Q is undergoing an independent AI risk assessment against [framework]. The certificate and report will be shared with every customer.

04

Data protection

Hosted on Google Cloud in Singapore. Our compliance programme maps to PDPA (Singapore), the Australian Privacy Act, PIPL (China), GDPR and the EU AI Act. SOC 2 is planned.

People in the loop

Admission decisions stay with your institution.

Q prioritises and recommends. It never accepts, rejects or ranks an application for a decision. Advisors can take over any conversation, and every answer Q learns has a named approver.

Architecture

Each institution runs in its own boundary.

Candidate channels

  • Telegram
  • WhatsApp
  • Email
  • Web chat

Isolated per institution · Google Cloud, Singapore

Knowledge

Your approved programme data

Memory

One thread per candidate

Q

Reasoning, actions

Governance

Approvals, audit trail

Your institution

Advisor Desk
CRM · SIS

Open with candidates

Q tells candidates they are talking to an AI agent, and how to reach a person at any point. Candidates can ask what Q holds about them and request its deletion, in line with your privacy policy.

For your security review

We complete institutional security questionnaires as part of every pilot. On request:

  • Security overview
  • Data-flow diagram
  • Subprocessor list
  • Data retention policy
  • Data processing agreement
  • AI risk assessment report, once issued
[Before launch: confirm encryption, retention, subprocessor and model-training statements with the build partner.]