Sumyfi AI Usage Policy

When Sumyfi may use AI

• To summarize trends in your spending, saving, and budgeting activity.
• To answer natural-language questions about your financial dashboard.
• To generate personalized educational guidance, reminders, and planning suggestions.
• To improve clarity, not to make binding financial, lending, insurance, or tax decisions for you.

Data Minimization Principles

• Only send data strictly necessary to answer your current query.
• Never send full account histories, all transactions, or all account details by default.
• Prefer aggregated, categorized, or masked data whenever possible.

Prohibited Data

• Never send: account numbers, card numbers, government IDs, login credentials, exact internal user IDs, or unmasked personal addresses/phone numbers.

Sensitive Data Handling

• Mask or generalize merchant names (e.g., “Starbucks” → “Coffee Shop”).
• Round or bucket balances and amounts (e.g., $5,432.12 → “~$5.4K”).
• Remove or generalize account names (e.g., “TD Chequing 1234” → “Chequing Account”).
• Only include transaction details if your query specifically requires them.

Explicit User Consent

• You must be clearly informed that your financial data may be processed by AI providers to generate insights and responses.
• Consent is required before enabling AI features during signup or feature activation flows.
• If we expand AI functionality in a materially different way, we will update this page and the associated product disclosures.

Logging and Audit

• Never log or store raw AI prompts containing financial data.
• If logging is required for debugging, all sensitive data is redacted first.
• Maintain an audit log of: categories of data sent, intent/reason, timestamp, and anonymized user.

AI Providers and Data Transfers

• AI features may rely on carefully selected third-party model providers operating under Sumyfi instructions.
• Those providers should only receive the minimum context necessary to complete the requested task.
• Depending on the provider and hosting region, limited AI-processing data may be transferred across provincial, state, or national borders.
• Provider terms, retention controls, and security commitments may evolve over time, and Sumyfi will review material changes before continuing use.

Retention and Operational Handling

• We aim to avoid retaining raw financial prompts longer than operationally necessary to provide the feature, secure the service, or investigate abuse.
• Where product analytics or error monitoring is used, AI-related telemetry should be limited to redacted or non-sensitive metadata wherever possible.
• Retention periods may vary by provider, incident-response requirements, and applicable law, and those limits are reviewed during vendor and security assessments.

Important limitations

• AI outputs are informational and may be incomplete, generalized, or inaccurate.
• AI-generated responses are not legal, tax, accounting, credit, or investment advice.
• You remain responsible for reviewing your own financial decisions and account activity.

Human Review and User Rights

• AI features are designed to support users, not to make solely automated decisions that approve, deny, or materially restrict financial products or account access.
• If an AI-generated response seems wrong, incomplete, or risky, you should rely on human review and authoritative records before acting.
• You can stop using AI-powered features at any time, and you may contact Sumyfi support if you need help understanding how an AI feature handled your request.
• Additional privacy, access, correction, deletion, or complaint rights may be described in the Privacy Policy depending on your jurisdiction.

Review and Updates

• This policy is reviewed at least annually and after major product, vendor, or compliance changes.
• Continued use of AI-powered features after an update means the revised policy applies to those features.

What users should reasonably expect from AI inside a finance product

Financial software sits inside a higher-trust environment than many other consumer apps. That means users should expect more than convenience claims. They should expect clear disclosures, bounded use cases, and an explanation of how the product avoids sending unnecessary personal or financial information into AI workflows. Sumyfi treats that expectation as part of product design, not only as a legal requirement.

In practice, that means AI should help interpret financial context, summarize activity, and make the next step easier to understand. It should not quietly expand into unrelated decision-making or hide the fact that external providers may be involved. The goal is to make the system more understandable while preserving user trust.

How AI requests are scoped down before they are sent

When AI features are used, the request should be narrowed to the specific task the user is asking Sumyfi to help with. That can include summarizing a trend, comparing a small set of categories, or explaining why a pattern might matter. It should not require sending the full account profile, every transaction, or unrelated personal data simply because that data exists elsewhere in the product.

This scoping discipline matters for both privacy and quality. Smaller, better-bounded prompts reduce unnecessary exposure and also make it easier to produce clearer, more relevant output. In a finance setting, that tradeoff is usually better than maximizing raw model context.

Why this page exists as a public trust surface

Users, evaluators, partners, and search engines increasingly expect AI-enabled finance products to explain their operating boundaries publicly. A visible AI usage policy lowers uncertainty for people deciding whether the product is safe enough to explore, whether the guidance is meant to be educational rather than authoritative, and whether the company is treating AI as a controlled tool instead of a vague marketing layer.

Sumyfi publishes this page so users can review those boundaries before relying on AI-powered features. It works alongside the Privacy Policy, Terms of Service, and security materials to create a fuller picture of how the product handles sensitive contexts.

What AI should feel like inside a finance workflow

In a budgeting or money-management product, AI should feel assistive rather than invasive. The useful version of AI helps a user understand trends faster, recognize category drift sooner, and turn a crowded dashboard into a smaller number of practical next steps. It should reduce interpretation effort. It should not create new uncertainty about what data was used, why a recommendation appeared, or whether the system is acting beyond the user request.

Sumyfi treats that user expectation as part of product quality. A finance user is not simply looking for novelty. They are looking for a system that helps them review important information with less confusion and less hidden risk. That is why this page focuses on operational boundaries, not only marketing claims about smart features.

How this policy supports evaluators, compliance review, and user confidence

Public AI usage documentation serves more than one audience. Users need it so they can judge whether an AI-powered feature is appropriate for their own comfort level. Internal teams and external reviewers need it so there is a stable reference point for what the product is expected to do, what kinds of data handling are in bounds, and where escalation or redesign should happen if the implementation shifts over time.

For that reason, Sumyfi treats this page as a living product-policy surface. It helps connect user expectations, privacy commitments, vendor review, and feature design into one readable explanation. That makes the AI layer easier to evaluate and easier to challenge when a proposed use case would stretch beyond what belongs in a high-trust financial environment.