Raffi uses the question you send in Teams to find and generate an answer from SRMC-approved information. Common identifiers are removed before AI processing and transcript storage. Pseudonymised transcripts are normally deleted after 30 days, and Raffi does not use student conversations to train a general-purpose AI model.
1. Overview and roles
This Notice supplements SRMC’s Data Protection Notice and explains the information handling specific to Raffi.
“We”, “us” and “our” in this Notice refer to coai when describing operation of Raffi. Questions about your wider student record or SRMC’s purposes should be directed to SRMC.
2. Scope
This Notice applies when an authorised student interacts with Raffi in Microsoft Teams, uses its feedback controls, or is referred by Raffi to an authorised SRMC staff member. It also covers the limited operational data used to secure and maintain the service.
It does not replace the privacy terms for Microsoft Teams, SRMC’s Microsoft 365 environment, SRMC’s websites or other SRMC systems. Those are governed by SRMC’s arrangements and notices.
3. Information Raffi handles
Information you provide
- the questions, follow-up messages and other text you send to Raffi;
- optional “helpful” or “not helpful” feedback; and
- information you include in a conversation even if Raffi did not ask for it.
Information generated during use
- Raffi’s answer, cited source names and links, whether an answer was available, language, response time and similar service information;
- pseudonymous user and conversation references created using a one-way keyed process, with no lookup table that maps the stored reference back to you;
- a short-lived, identifier-scrubbed conversation context so Raffi can understand a follow-up question; and
- security and operational records such as event type, status, timing and conversation type. These logs are designed not to contain message text, names, directory identifiers, authentication credentials or secrets.
Information provided by SRMC
Raffi searches passages from the specific SharePoint knowledge library approved by SRMC. The library may contain personal data only to the extent SRMC includes it in those approved documents.
Information Raffi is not designed to collect
Raffi is not designed to accept or read file, image or voice attachments, payment information, or records from SRMC’s student-records systems. Please do not send national identification or passport numbers, passwords, payment-card or bank details, medical or counselling records, private contact details, or another person’s confidential information.
4. Why the information is used
Raffi processes information only as reasonably needed for SRMC to:
- understand a question, including translating it for information retrieval when necessary;
- search the approved SRMC knowledge library and generate an answer with citations;
- maintain short conversation context for follow-up questions;
- apply safety controls and refer certain matters to an authorised staff member where that feature is enabled;
- record feedback, evaluate answer quality and improve Raffi for SRMC;
- produce aggregate service reports that do not identify students;
- prevent misuse, maintain security, troubleshoot faults and monitor reliability; and
- comply with applicable law, enforce relevant terms and respond to a lawful request.
coai does not sell student personal data, use it for advertising, build advertising profiles, or use student conversations to train a general-purpose AI model.
5. Automated processing and identifier removal
Raffi uses automated language and search systems to understand messages, retrieve relevant SRMC passages and generate an answer. It does not make academic, admissions, disciplinary, financial, immigration, health, legal or other consequential decisions about a student.
Before a question is sent for AI text processing or written to Raffi’s transcript store, Raffi automatically replaces common patterns resembling national identifiers, passport numbers, email addresses, telephone numbers and long number sequences. Answers are checked through the same process before transcript storage.
Redaction reduces risk; it is not a guarantee. A name, address or unusual identifier may not be detected. The safest approach is not to include personal or sensitive information in the first place.
If Raffi cannot find a sufficiently relevant approved source, it is designed to say so rather than invent an answer. Important information must still be verified with official SRMC materials or staff.
7. Location and overseas transfers
coai configures Raffi’s application hosting, operational database and AI text processing in Singapore. Raffi does not intentionally transfer student message content outside Singapore as part of those service operations.
You access Raffi through SRMC’s Microsoft 365 environment. The location and handling of information within Teams, SharePoint, Microsoft Graph and related Microsoft services are governed by SRMC’s Microsoft configuration and agreements. Accessing Raffi from outside Singapore also causes information to travel to and from your device.
If a future service change would involve an overseas transfer of personal data, SRMC and coai will apply the safeguards required by the PDPA, including a standard of protection comparable to the PDPA, and update this Notice where the change is material.
8. How long information is kept
- Conversation context: identifier-scrubbed recent exchanges are held in memory for up to 30 minutes and are not written to disk as conversation memory.
- Conversation transcripts and feedback: identifier-scrubbed questions and answers, pseudonymous references and associated feedback are normally deleted automatically after 30 days.
- Escalation mapping: if staff escalation is enabled, the pseudonymous mapping used to reuse an escalation chat is normally deleted after 30 days. Messages in the resulting Teams chat are retained under SRMC’s Microsoft 365 policies.
- Operational records: non-content security and reliability records are kept only for the applicable operational, contractual or legal period and are then deleted or de-identified.
- Knowledge-library material: SRMC controls the source documents and their retention. Raffi retains processed library passages only while needed to provide the service or as directed by SRMC.
Information may be retained for a longer period where necessary to investigate a security incident, comply with law, resolve a dispute or preserve evidence. In that case it will be limited to the material needed and protected until it can be deleted or de-identified.
9. How information is protected
Raffi uses administrative, technical and organisational safeguards proportionate to the information handled. These include authenticated single-tenant access, encrypted network connections, least-privilege access to SRMC’s approved knowledge library, pseudonymous stored identifiers, automatic identifier scrubbing, restricted database access, service monitoring, rate limits, retention controls and incident-response procedures.
No system can be guaranteed completely secure. Students can reduce risk by using Raffi only through their authorised SRMC account and by not submitting unnecessary personal or confidential information.
10. Your choices and rights
Subject to the PDPA and applicable exceptions, you may ask for access to or correction of personal data, ask how it was used or disclosed, withdraw consent where processing relies on consent, or raise a concern about its handling.
Because SRMC determines the purposes of Raffi’s processing and coai acts as its data intermediary, please send these requests to the SRMC Data Protection Officer at DPO@srmc.edu.sg with “SRMC DPO – Raffi” in the subject line. coai will assist SRMC as required. We may need information to verify your identity and locate the relevant records.
Withdrawing consent or objecting to processing needed to operate Raffi may mean you can no longer use Raffi, but SRMC’s other student-support channels remain available. You may also contact hello@coai.sg about Raffi-specific privacy or security concerns.
11. Younger users
Some students may be under 21. Raffi is available only through the SRMC environment, uses information for the limited student-support purposes described above, avoids advertising and behavioural profiling, and is designed to minimise stored identifiers. Where consent or authorisation from a parent or guardian is required, SRMC is responsible for arranging it.
Students should ask a parent, guardian or SRMC staff member for help if they do not understand this Notice or are unsure whether to share information.
12. Personal data breaches
coai maintains procedures to identify, contain and investigate suspected personal data breaches affecting Raffi and to notify SRMC without undue delay. SRMC assesses whether notification to the Personal Data Protection Commission or affected individuals is required and makes those notifications in accordance with the PDPA. We will support SRMC’s investigation and response.
13. Updates to this Notice
We may update this Notice when Raffi’s information handling, law or contractual arrangements change. The effective date at the top shows the latest version. Material changes will be communicated through this page, Raffi or an appropriate SRMC channel before or when they take effect, as required.
Privacy contacts
SRMC Data Protection Officer: DPO@srmc.edu.sg
Raffi operator enquiries: hello@coai.sg
For general student support, use SRMC’s usual Student Services channel.