AI Use Disclosure

Effective date: to be set on adoption.

1. The Short Version

Your Pocket Planner uses artificial intelligence (AI) to help you build and understand your financial plan. We want you to know what that means in practice, what AI does, what it does not do, and what choices you have. The points below are the most important. The remaining sections give the detail.

• AI helps generate explanations and projections inside your personalized plan, answers your questions about your situation in plain language, and extracts data from documents you upload (with your consent).
• AI does NOT recommend specific securities, mutual funds, ETFs, insurance products, or any other financial product. AI does NOT give trade instructions. AI does NOT manage your money. AI does NOT replace a regulated professional.
• AI can make mistakes. Outputs can be inaccurate, out of date, or misaligned with your specific situation. You should treat AI outputs as a starting point, not as conclusions.
• Changes the AI proposes to your plan are surfaced to you for confirmation before they are applied.
• We send your messages and relevant plan information to a small number of third-party AI providers (Section 5). They are contractually prohibited from using your information to train their models.
• You can opt out of optional AI features. Some AI use is necessary to provide the Service; opting out of those features means the Service cannot be delivered.

2. Why We Use AI

Personalized financial planning is information-dense. Most Canadians do not have access to affordable, individualized planning help. AI lets us produce explanations, calculations, and tailored guidance at a scale that would otherwise require many hours of one-to-one professional time. It also lets us answer your follow-up questions in plain language at any time, instead of leaving you with a static document.

AI is one ingredient in the Service. The other ingredients are: a Canadian planning calculation engine (deterministic, not AI), a curated source-of-truth knowledge base maintained by our team, and human review of the Service's outputs through ongoing monitoring.

3. What AI Does in the Service

3.1 The planning assistant (N.I.E.L.)

N.I.E.L. is our conversational planning assistant. When you ask it a question, it uses AI to:

• Understand what you are asking.
• Use the information you have already entered into your plan, plus the curated planning knowledge base we maintain, to draft a response.
• Identify whether your message contains information that should update your plan (for example, a new income figure, a debt paid off, or a change in employment), and if so, propose a structured update to your plan.
• Present the proposed update for your explicit confirmation before any change is saved.

Where N.I.E.L. is uncertain, it is instructed to say so rather than guess, to refer you to a qualified professional where appropriate, and to refuse questions outside its scope.

3.2 Document extraction

If you upload a document (for example, a pension statement or Notice of Assessment), AI is used to extract the data points relevant to your plan. Extracted values are presented to you for confirmation before they are written to your plan. You can correct or reject any extracted value.

3.3 Plan generation and projections

Numeric projections (retirement income, drawdown sequences, tax estimates) are produced by deterministic calculations, not AI. AI is used to write the plain-language summaries, explanations, and tailored commentary that appear alongside those numbers.

3.4 Educational content

Some educational content in the Service (for example, explanations of tax concepts or government benefits) is reviewed by our team but drafted with AI assistance. Where regulatory content sensitivity is high, we apply additional human review before publication.

3.5 Voice and other modalities

We offer optional voice-based interaction with the planning assistant, including a spoken Discovery conversation. When you use it, audio of your voice is processed by Google through the paid Gemini API to recognize speech and to generate spoken responses (see Section 5). Voice is opt-in and separately consentable; you can use the Service without it.

4. What AI Does NOT Do

These limits are central to what the Service is. They are enforced through the design of the AI assistant, through the prompts it operates under, and through our monitoring.

• The Service does not recommend specific securities, mutual funds, ETFs, annuities, segregated funds, insurance products, or other financial products. The AI is instructed to refuse such requests.
• The Service does not provide instructions to buy, sell, or hold a specific investment.
• The Service does not act as your investment advisor, securities dealer, mutual fund dealer, exempt market dealer, portfolio manager, insurance agent, mortgage broker, lawyer, accountant, or tax preparer.
• The Service does not assess whether a specific product is suitable for you under National Instrument 31-103 or any other regulatory suitability standard.
• The Service does not promise any particular return, tax outcome, government benefit, or other financial result.
• The Service does not make fully automated decisions that produce a legal or similarly significant effect on you. AI-proposed plan updates and extracted document values are presented to you for explicit confirmation before they are applied.

5. Which AI Providers We Use, and What Goes to Them

We do not run our own foundation models. We use a small number of third-party AI providers, listed below. Each provider is bound by a written agreement that, at minimum:

• Permits the use of your information only for the purpose of generating responses, transcriptions, or extractions for the Service.
• Prohibits the use of your information to train, fine-tune, or otherwise improve their generally available models.
• Requires confidentiality and security safeguards proportional to the sensitivity of the data.
• Requires notification to us in the event of a security incident.

5.1 Current providers

• Anthropic, PBC (United States). Powers the N.I.E.L. planning assistant and document extraction. When you chat with the assistant or upload a document, your messages and the relevant context from your plan are sent to Anthropic to generate a response. Anthropic processes your information in accordance with its Commercial Terms and Privacy Policy and is contractually prohibited from using your information to train its models.
• Google LLC (United States). Powers the optional voice features of the planning assistant, including the spoken Discovery conversation. When you use voice, the audio and the relevant context from your plan are sent to Google through the paid Gemini API to recognize speech and generate spoken responses. Under the paid Gemini API, Google does not use your prompts or responses to train or improve its models, acts as a data processor, and retains this data only briefly to detect abuse and meet legal or regulatory requirements.
• Additional providers may be introduced in the future, for example a specialized optical character recognition (OCR) provider for document handling. This Disclosure will be updated before any new provider begins processing your personal information.

5.2 What we send to AI providers

We send only what is necessary for the AI to perform its task:

• Your messages to the planning assistant and the assistant's prior responses in that conversation.
• Relevant pieces of your plan that the assistant needs to answer the question (we do not blindly send your entire plan).
• Curated reference material from our internal knowledge base that helps the assistant answer accurately and consistently.
• For document extraction: the contents of the document you have uploaded, along with the specific fields we are asking the AI to extract.

We do not send your password, your payment card number, or other credentials to AI providers. Where feasible, we redact or tokenize highly sensitive identifiers (for example, a Social Insurance Number that has been entered against our guidance) before the information is sent to an AI provider.

5.3 Cross-border processing

Our current AI providers process information in the United States. Your information is therefore subject to U.S. law while it is being processed by these providers, including the possibility of lawful access by U.S. government authorities (for example, under the U.S. CLOUD Act). Our Privacy Policy and the section above describe the safeguards we apply to these transfers.

6. We Do Not Train Models on Your Information

We do not use your personal information to train, fine-tune, or improve any AI model that we or our vendors offer. Our agreements with AI vendors expressly prohibit such use of your information.

Where we improve the Service by analyzing how clients use it, we do so using aggregated or de-identified information, and any model improvement work we conduct internally uses synthetic test data or consented samples reviewed by our Privacy Officer. If we ever propose to train a model on personal information, we will obtain your separate, express consent in advance.

7. Limits, Errors, and How to Treat AI Outputs

AI is useful, but not infallible. You should know its main limitations before relying on what it says.

7.1 Hallucinations

AI models can produce confident-sounding statements that are factually incorrect. We mitigate this by grounding the assistant in our curated knowledge base, by limiting the topics it answers on, and by instructing it to defer to qualified professionals or to say it does not know. Despite these measures, occasional errors are possible. Treat the assistant's outputs as a starting point, not as conclusions.

7.2 Sensitivity to the inputs you provide

The plan and the responses are only as accurate as the information you have entered. Missing, outdated, or incorrect information will produce missing, outdated, or incorrect outputs.

7.3 Currency of information

Canadian tax rules, benefits, contribution limits, and other rules change. We maintain a documented refresh process for the planning knowledge base and the underlying calculation engine. Even so, there will be moments when a change in law is not yet reflected. Where currency is uncertain on a specific question, the assistant should tell you so.

7.4 Out-of-scope questions

The assistant is instructed to decline questions outside its scope, including legal advice, medical questions, instructions to buy or sell specific securities, and questions that would require a regulated professional licence to answer in Canada. If you receive a response that appears to step over one of these limits, please tell us so we can review and improve our controls (Section 10).

8. How We Log and Review AI Use

We log the assistant's conversations with you, the structured updates it proposes, the documents you upload, and the AI's outputs. Logs are treated as Sensitive personal information in our Privacy Governance Program. We use logs to:

• Show you your own conversation history when you return to the Service.
• Investigate specific incidents (for example, a reported error or a support ticket).
• Sample outputs internally for quality and safety review.
• Retrain our prompts and refine our knowledge base, not the underlying models.

You can delete individual conversations from your account. See our Privacy Policy (Sections 8 and 10) for the retention and deletion rules.

9. Your Choices

9.1 AI is required for parts of the Service

AI is essential to certain features of the Service, including the planning assistant and the conversational explanation of your plan. You cannot use those features without AI processing. If you do not wish to use AI, you can still see your plan figures and the basic deterministic calculations, but the conversational interface and document extraction will not be available.

9.2 Optional AI features

Some AI features (for example, voice interaction when launched) are optional and separately consentable. You can opt in or out of these in your account settings.

9.3 Deleting AI-related content

You can delete individual conversations from your account at any time. Deleting your account deletes the associated conversation history in accordance with our Privacy Policy.

10. Telling Us When AI Gets It Wrong

If you notice an AI output that is inaccurate, inappropriate, or that appears to step over the limits in Section 4, please tell us:

• Use the in-product feedback control next to the AI response, where available.
• Or email hello@yourpocketplanner.ca with a description of what happened and (if you are comfortable) a screenshot or link to the conversation.

We review feedback as part of our quality and safety process. Where feedback indicates a systemic issue, we update prompts, knowledge base content, or training materials accordingly. We do not retaliate against users who give us feedback.

11. Changes to This Disclosure

Artificial intelligence and the rules around it are evolving quickly. We will update this Disclosure whenever we change how AI is used in the Service in a way that affects what is described above. Material changes are communicated to active users as described in our Privacy Policy. The version and effective date are shown at the top of this Disclosure.

12. Contact Us

For questions about how AI is used in the Service, contact us at:

• Privacy questions: hello@yourpocketplanner.ca
• Product or AI behaviour questions: hello@yourpocketplanner.ca
• Mailing address: to be inserted on adoption (counsel-coordinated)

Effective date: TO BE SET ON ADOPTION

DRAFT v0.1 . Authored 2026-05-12 . Pending counsel review.