Alpha AI Copilot Terms of Use
Last updated: May 2026 | Version 2.0
Includes the AI Skills terms (Estate Planning, Retirement Planning, and future Skills).
1. Definitions
In these Terms of Use (“Terms”):
- “Company”, “we”, “us”, or “our” means Sybil Pte Ltd.
- “Alpha” or “AI Copilot” means the AI-powered assistant features in the QuantQuest platform, branded as “Alpha”, including the Engagement Copilot, Client Copilot, Insurance Analysis, Knowledge Pack, and related AI-assisted tools.
- “Adviser”, “you”, or “your” means the financial adviser or representative using the AI Copilot through their QuantQuest account.
- “Client Data” means personal data and financial information of your clients processed by the AI Copilot, including names, financial records, insurance policies, net worth data, and uploaded documents.
- “AI Output” means any text, analysis, summary, draft, or other content generated by the AI Copilot.
- “LLM” means the third-party large language model used to generate AI Outputs.
- “Knowledge Pack” means the adviser-curated library of product reference material you upload for the AI Copilot to reference.
2. Introduction
These Terms govern your use of the AI Copilot. They supplement the QuantQuest Platform Terms of Service. Where these Terms conflict with the Platform Terms with respect to AI Copilot use, these Terms prevail.
Our AI Copilot architecture is aligned with Singapore's Personal Data Protection Act 2012 (PDPA) and the Monetary Authority of Singapore's FEAT Principles (Fairness, Ethics, Accountability, Transparency).
3. Nature of the AI Copilot
The AI Copilot is a technology tool that helps financial advisers organise data, analyse information, and manage client work. You acknowledge:
- Sybil Pte Ltd is a technology company, not a licensed financial advisory firm.
- The AI Copilot is a decision-support tool. It does not provide financial advice, product recommendations, investment opinions, or suitability assessments.
- AI Outputs may contain errors, inaccuracies, hallucinations, or outdated information.
- The AI Copilot organises and summarises data you and your clients enter. It does not independently verify that data.
- You must independently verify AI Outputs before relying on them or sharing them with clients.
- The AI Copilot does not replace your professional judgment, training, or regulatory obligations.
4. Professional Responsibility
You acknowledge:
- You retain full professional and regulatory responsibility for all advice, recommendations, and decisions made for your clients, regardless of whether AI Outputs informed those decisions.
- You bear sole responsibility under Financial Advisers Act section 36 and Regulation 18B for forming the reasonable basis for any advice or recommendation given to your clients, regardless of whether AI Outputs informed your analysis.
- You are responsible for critically evaluating AI Outputs and applying independent professional judgment before acting on them.
- You will not present AI Outputs to clients as your finalised professional analysis without first reviewing and adapting them.
- Your use of the AI Copilot does not reduce your obligations under the FAA, the Financial Advisers Regulations, the MAS Guidelines on Fair Dealing, or any other applicable laws and regulations.
- In line with MAS FEAT Principle 8, you remain accountable for AI Outputs whether the AI was developed internally or by a third party.
5. AI Copilot Operating Modes
QuantQuest operates in two modes. No other operating mode exists or can be activated.
- (a) Mode A — General Analysis. Where no Knowledge Pack is active in the session, QuantQuest provides general financial analysis only. QuantQuest will refuse product-specific discussion and will not identify, compare, or comment on any specific investment product.
- (b) Mode B — Knowledge Pack Retrieval. Where a Knowledge Pack is active in the session, QuantQuest retrieves and surfaces content from the Knowledge Pack only in response to an explicit @-mention by you referencing a specific item in that Knowledge Pack. QuantQuest does not rank, shortlist, recommend, promote, fuzzy-match, or otherwise infer references from the Knowledge Pack. Every @-mention is logged against your session. An @-mention in one turn is not carried forward to subsequent turns; each reference must be re-stated.
Sybil does not provide, and does not purport to provide, financial advice or recommendations on any investment product within the meaning of the Financial Advisers Act 2001 or Notice FAA-N16. In all modes, AI Outputs are decision-support analysis only. You remain the licensed financial adviser and bear sole responsibility under FAA section 36 / Regulation 18B for any advice or recommendation communicated to your clients.
6. Data Processing
6.1 How Client Data Is Processed
When you use the AI Copilot:
- Client Data passes through automated identifier-masking before transmission to the inference layer. Across all paths, we mask NRIC numbers (with checksum validation), FIN numbers, Singapore mobile numbers, email addresses, bank account numbers, and Singapore 6-digit postal codes. For documents routed through the Knowledge Pack or insurance OCR pipelines, additional redaction applies to labelled personal fields (such as Life Assured, Date of Birth, Age / Age at Entry / ANB / ALB, Smoker Status, Gender, Occupation, Nationality, and policy identifiers), name references in prose (title-name patterns such as Mr, Mrs, Ms, Mdm, Dr), and labelled medical content (Medical History, Loading Reasons, Underwriting Remarks, Exclusion Reasons). A secondary detector flags any medical terms that survive redaction for audit review. No automated system is 100% effective; some personal data may be present in transmitted content. This masking is privacy-protective but does not constitute anonymisation under the PDPC Guide to Basic Anonymisation.
- All data is transmitted to the inference layer over encrypted connections (TLS 1.2 or 1.3).
- Client Data is not used to train, fine-tune, or improve any model. See Section 6.6 for our separate use of fully anonymised and aggregated data.
6.2 Singapore Region Deployment
The AI Copilot operates from a Singapore-region inference deployment, on infrastructure holding MTCS SS 584 Level 3 (covering the model inference service), ISO 27001 / 27017 / 27018 / 27701, and SOC 2 Type II certifications.
Inference compute may be dynamically scheduled across a global compute pool excluding mainland China to meet capacity and latency requirements. Where this transient cross-border processing occurs, it is governed by contractual data processing terms providing comparable protection to PDPA, supplemented by APEC Cross-Border Privacy Rules and Privacy Recognition for Processors certifications — both recognised mechanisms under Section 26 of the PDPA.
Certification evidence and infrastructure details are available to financial-adviser firms under a non-disclosure agreement to support due diligence and outsourcing risk assessment.
6.3 Data Handling at the Inference Layer
The AI Copilot is architected with the following data-handling guarantees at the inference layer:
- Prompts and outputs (Client Data transmitted for inference) are not retained.
- Only de-identified call-status metadata (timestamps, token counts, success/failure codes) is logged for operational purposes.
- Client Data is not used for model training, fine-tuning, or improvement of any model.
- Cache content used to optimise inference performance is isolated per-customer and is not accessible to or shared with other parties.
6.4 QuantQuest-Side Conversation History
Separately from the inference-layer zero-retention posture above, QuantQuest stores conversation history (your messages and AI responses) on QuantQuest servers in Singapore to enable session continuity, audit traceability, and your ability to review prior sessions. This data is:
- Stored encrypted at rest (AES-256) on Singapore-region database infrastructure.
- Strictly isolated per-adviser. Tenancy filters are enforced at the database read layer on every query. No other adviser, and no client, can access your conversation history.
- Retained for as long as your account is active, or as required by applicable law.
- Deletable by you at any time through the AI Copilot interface, individually by session.
- Deleted within 90 days of account termination, unless retention is required by law.
6.5 Data Boundaries
For transparency:
- The AI Copilot does not have access to NRIC numbers in unmasked form, bank account credentials, login passwords, or payment card details.
- Each adviser's data is strictly isolated. The AI Copilot cannot access data belonging to other advisers' clients. This isolation is enforced at the database layer on every read.
- The Engagement Copilot (practice-level) has access only to client names and engagement scores. It does not access detailed financial data, insurance policies, or personal records.
- Uploaded documents are processed in isolated sessions and are not shared across clients or advisers.
6.6 Aggregated and Anonymised Data
We may use fully anonymised and aggregated data — with all identifiers removed in accordance with the PDPC Guide to Basic Anonymisation — for purposes including product improvement, research, internal model evaluation, and business operations. Such anonymised data is not personal data under the PDPA. If we materially change how we use this data in a way that affects you, we will update these Terms and require re-acceptance.
7. PDPA Compliance
With respect to the Personal Data Protection Act 2012 (“PDPA”):
- Sybil Pte Ltd acts as a data intermediary processing Client Data on your behalf for the purpose of providing the AI Copilot service. Our processing relationship with you is governed by these Terms and the QuantQuest Data Protection Policy.
- You are responsible for obtaining the necessary consents from your clients for the processing of their personal data through AI services.
- You must inform your clients that AI-assisted technology supports your work, in line with MAS FEAT Principle 12 (proactive disclosure of AI use).
- We recommend including AI processing disclosure in your client engagement agreements or fact-find consent forms. Template language is available on request.
- You must promptly notify us of any client access, correction, or consent-withdrawal request affecting data processed by the AI Copilot.
- Where Client Data may be processed outside Singapore through transient inference compute scheduling (see Section 6.2), the transfer mechanism is the contractual data processing terms governing the inference layer and APEC CBPR/PRP certifications — both recognised under PDPC's Section 26 transfer guidance. We will provide evidence of these mechanisms to your firm on request.
8. Data Breach Notification
If we become aware of a data breach affecting Client Data:
- We will notify you without undue delay upon confirming the breach affects Client Data within your account. Our operational target is notification within 24 hours of confirmation; where there is credible indication of a high-severity breach, we target initial notification within 1 to 4 hours to give you maximum runway on your downstream regulatory obligations.
- We will provide sufficient information for you to assess the breach and meet your own notification obligations under the PDPA (PDPC notification within 3 calendar days of your assessment of notifiability under Part 6A) and under MAS Notice FSM-N23 (MAS notification within 1 hour of discovery of relevant cyber incidents).
- You acknowledge that certain categories of Client Data — including life, accident, and health insurance information, and medical information — are prescribed under the PDPA Breach Notification Regulations as triggering automatic mandatory PDPC notification if involved in a breach with identifiable individuals.
- Where the underlying incident originates with an upstream infrastructure provider, that provider is contractually obliged to notify us without undue delay; we will pass that notification through to you on the operational timeline above. We do not artificially gate notification on receiving a final root-cause report from upstream.
- You are responsible for maintaining your own breach response plan and notifying affected clients, the PDPC, and MAS as required.
9. Acceptable Use
You will NOT:
- Treat AI Outputs as a substitute for your own professional assessment of client needs and suitability.
- Present AI Outputs to clients as finalised professional advice without prior review and adaptation.
- Enter NRIC numbers, FIN numbers, passport numbers, birth certificate numbers, bank account numbers, or credit/debit card numbers into the AI Copilot. These identifiers are not required for AI analysis.
- Enter personal data of individuals who have not consented to AI processing.
- Attempt to extract, reverse-engineer, or manipulate the AI system through prompt injection, jailbreaking, or other adversarial techniques.
- Use the AI Copilot to generate content that is misleading, discriminatory, or unlawful.
- Rely solely on AI-generated coverage analysis, gap assessments, or financial summaries without independent verification.
- Use the AI Copilot for purposes outside your financial advisory practice.
10. Insurance Document Processing (OCR)
The AI Copilot can process uploaded insurance documents via OCR and AI-assisted data extraction. You acknowledge:
- Uploaded documents may contain sensitive information including medical history, health declarations, and policy exclusions.
- You must obtain client consent before uploading their insurance documents for AI processing.
- OCR processing is performed in our Singapore-region document processing infrastructure with PII scrubbing applied before any content is forwarded to the inference layer.
- Extracted data should be reviewed for accuracy before being saved. AI extraction may misread, omit, or misinterpret content, particularly with respect to monetary values, premium tables, benefit schedules, and exclusion text.
- Documents are not retained at the inference layer after processing per Section 6.3.
11. Knowledge Pack
The Knowledge Pack architecture separates content into two scopes:
- Library scope: Generic, non-client-specific reference material such as insurer brochures, product summaries, and adviser notes. Library content is per-adviser scoped and never shared across advisers.
- Client-vault scope: Client-specific documents such as benefit illustrations, fact-finds, or policy schedules attached to a named client record. Client-vault content is bound to that specific client record and is not accessible from sessions involving any other client.
You acknowledge:
- For client-vault uploads, the named client must have consented to AI-assisted processing under your firm's documentation (see Section 7). Each upload must be scoped to a single named client.
- An automated personalisation classifier runs on every upload. Documents flagged as containing client-specific data are rejected from the library scope and must be uploaded under a client-vault scope.
- PII scrubbing (per Section 6.1) is applied to all Knowledge Pack content before any content is forwarded to the inference layer.
- As with any AI Output, you remain accountable for your Financial Advisers Act section 36 reasonable-basis obligation (see Section 4). Verify factual and numerical content from Knowledge Pack analyses before using it with a client.
- AI Outputs that reference Knowledge Pack content may contain extraction errors, including OCR misreads of monetary values, benefits, premium tables, or exclusion text. Verify all numerical and benefit-related statements against original source documents before client communication.
- A SHA-256 audit hash is recorded with each AI response that references your Knowledge Pack, enabling forensic reconstruction of the exact reference state at generation time.
- You may delete Knowledge Pack entries at any time. Deletion removes the entry from future AI access; existing AI Outputs that previously referenced the entry retain their citations for audit purposes.
12. Citations and Source Attribution
AI Outputs that draw on retrievable sources (Knowledge Pack content, structured client records, or calculator tools) are accompanied by “Sourced” attribution badges identifying the underlying reference. The Sourced badge is a traceability mechanism: it indicates the AI Output is grounded in a specific identifiable source available within your account, not a correctness or completeness guarantee. You remain responsible for verifying the substance of any AI Output, including any numerical, factual, or benefit-related statements derived from cited sources.
13. AI Output Disclaimers
All AI Outputs in the AI Copilot are accompanied by persistent disclaimers. You agree:
- These disclaimers form part of these Terms.
- Removal or circumvention of these disclaimers is prohibited.
- If you share or export AI Outputs (in reports, client communications, or otherwise), you should include appropriate caveats about the AI-assisted nature of the content.
14. Intellectual Property
All platform features, AI models, prompts, safety systems, and other technology provided through the AI Copilot remain the intellectual property of Sybil Pte Ltd. You are granted a limited, non-exclusive, non-transferable licence to use the AI Copilot solely to support your financial advisory practice through the QuantQuest platform. Client Data you input remains yours. AI Outputs generated from your Client Data may be used by you without restriction, subject to the accuracy caveats in these Terms.
15. Disclaimers and Limitation of Liability
15.1 No Warranty
THE AI COPILOT AND ALL AI OUTPUTS ARE PROVIDED ON AN “AS IS” AND “AS AVAILABLE” BASIS. WE MAKE NO WARRANTIES, EXPRESS OR IMPLIED, REGARDING THE ACCURACY, COMPLETENESS, RELIABILITY, OR SUITABILITY OF AI OUTPUTS FOR ANY PURPOSE. WE DO NOT GUARANTEE THE AI COPILOT WILL BE UNINTERRUPTED, ERROR-FREE, OR FREE FROM HARMFUL COMPONENTS.
15.2 Limitation of Liability
TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, SYBIL PTE LTD SHALL NOT BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, OR PUNITIVE DAMAGES ARISING FROM YOUR USE OF THE AI COPILOT, INCLUDING DAMAGES ARISING FROM RELIANCE ON AI OUTPUTS, ERRORS IN AI-GENERATED ANALYSIS, OR DECISIONS MADE BASED ON AI OUTPUTS.
OUR TOTAL AGGREGATE LIABILITY FOR ALL CLAIMS ARISING FROM OR RELATED TO THE AI COPILOT SHALL NOT EXCEED THE TOTAL AMOUNT PAID BY YOU FOR YOUR QUANTQUEST SUBSCRIPTION IN THE TWELVE (12) MONTHS IMMEDIATELY PRECEDING THE EVENT GIVING RISE TO THE CLAIM.
The limitations in this section do not apply to: (a) liability arising from death or personal injury caused by negligence; (b) fraud or fraudulent misrepresentation; or (c) any liability that cannot be excluded or limited under applicable law.
16. Indemnification
You agree to indemnify, defend, and hold harmless Sybil Pte Ltd, its officers, directors, employees, and agents from any claims, damages, losses, liabilities, costs, and expenses (including legal fees) arising from:
- Your use of the AI Copilot and reliance on AI Outputs.
- Your violation of these Terms.
- Any financial advice, recommendations, or decisions you make using AI Outputs.
- Any claim by your clients related to AI-assisted analysis or outputs.
- Your failure to obtain necessary client consents for AI data processing.
- Any breach of FAA section 36 / Regulation 18B obligations.
- Any regulatory action arising from your use of AI Outputs in your advisory practice.
17. Modifications to Terms
We may modify these Terms at any time. Material changes will require you to re-accept the updated Terms through the AI Copilot consent flow before continuing to use the service. The version number and effective date will be updated. Your acceptance of updated Terms is recorded with a new timestamp and version number.
18. Termination
We may suspend or restrict your access to the AI Copilot if you breach these Terms. You may stop using the AI Copilot at any time by disabling it through your account settings (Smart Assistant toggle). Termination of your QuantQuest subscription terminates your access to the AI Copilot. On termination, conversation history is handled in accordance with Section 6.4.
19. Governing Law and Dispute Resolution
These Terms are governed by the laws of the Republic of Singapore. Any dispute arising out of or in connection with these Terms is subject to the exclusive jurisdiction of the courts of Singapore.
20. Contact
For questions about these Terms or the AI Copilot:
Email: support@quantquest.sg
DPO: dpo@quantquest.sg
Company: Sybil Pte Ltd
21. Acknowledgment
By accepting these Terms through the AI Copilot consent flow, you acknowledge you have read, understood, and agree to be bound by these Terms of Use. You further acknowledge that the AI Copilot is a technology tool that does not provide financial advice, that you retain full professional responsibility for all client-facing advice and decisions, and that you will ensure compliance with the PDPA and all applicable regulations when using AI Outputs.
22. AI Skills
An “AI Skill” is a guided, multi-turn Alpha workflow that you switch on inside AI chat to work through a specific planning area and produce one or more deliverables. Estate Planning and Retirement Planning are the first Skills; others will be added over time and are all governed by this section. You accept these AI Skills terms once, the first time you use a Skill, and your acceptance is recorded separately so that using Skills never changes your acceptance of the rest of these Terms. You acknowledge:
- Every Skill deliverable is a draft and a working aid, not a finished professional product. You must review, verify and adapt it before relying on it or sharing it with a client, and you remain the licensed adviser who executes any advice.
- A Skill does not perform any regulated activity, exercise professional judgment, or carry out independent verification on your behalf. You retain full responsibility under the Financial Advisers Act for any advice informed by a Skill (see Section 4).
22.1 Estate Planning Skill
Sybil is a technology company. It is not a law practice and does not provide legal advice, and no solicitor-client relationship arises through this Skill. You acknowledge:
- The Skill can assemble a simple will draft for a civil (non-Muslim) Singapore estate from a fixed clause bank. It is a draft for your review and for vetting by a qualified Singapore lawyer, not an execution-ready will; all identity fields are left blank by construction.
- Whether any will is valid depends entirely on proper execution and the testator's capacity, which the platform does not supervise. The Wills Act requires signing before two witnesses present together, a testator aged at least 21, a beneficiary or a beneficiary's spouse not acting as a witness, and no alteration after signing. You must not represent a will produced through this Skill as valid or guaranteed.
- The Skill routes certain cases out to a qualified lawyer instead of drafting, for example Muslim estates, foreign immovable property, special-needs dependants, and contested circumstances. This is a design feature, not a guarantee that every unsuitable case is detected; you remain responsible for identifying cases that need a lawyer.
- Some assets do not pass under a will and are not addressed by the draft, including CPF savings, insurance moneys subject to a nomination under the Insurance Act, and property held in joint tenancy. The draft does not make, alter or revoke any CPF or insurance nomination. Any client-facing estate document carries a persistent “not legal advice” disclaimer you must not remove.
22.2 Retirement Planning Skill
- Every retirement figure, projection and CPF LIFE payout the Skill shows is an illustration or estimate, not a guarantee, and must be verified against the client's actual saved plan. CPF LIFE payouts are non-guaranteed ranges that vary with interest rates and life expectancy and must not be presented as promised amounts.
- Any illustrated investment return is subject to the MAS illustration ceiling (currently 8% per annum), is not a promise of performance, and investment values can fall as well as rise.
- The Skill can compose an investment-linked policy (ILP) funding proposal and create it as a draft portfolio. When it does, you remain the adviser of record, form your own reasonable basis for any recommendation, and confirm its suitability under the Financial Advisers Act and the Financial Advisers Regulations before it reaches the client.
- The Skill can save a proposal onto the client's goals chart, create a draft ILP portfolio, and pre-fill the CPF planner. These are writes and hand-offs into the client's live records that you review, and that you, not Alpha, execute before anything reaches the client.
22.3 Data, liability and updates
The data-processing, PDPA, limitation-of-liability and indemnity terms elsewhere in these Terms apply in full to Skills. You remain responsible for obtaining the client consents required under the PDPA before entering personal data, including any third-party data such as a named executor, guardian or beneficiary, into a Skill. We may update this AI Skills section; a material change requires you to accept the updated Skills terms before continuing to use Skills, recorded separately from your acceptance of the rest of these Terms.
Document Version: 2.0 | Effective Date: May 2026 | Sybil Pte Ltd. All rights reserved.