🔒
Finance Clarity
Enter the passcode to view this presentation
Incorrect passcode
01 / 11
Passport To Wealth × Finance Site Generator

Personalized Financial Clarity for Every Client

A product vision for turning fragmented bank data into beautiful, shareable financial reports — without the compliance burden of handling client data.

Prepared for Arielle Tucker, CFP®, EA April 2026 Confidential
02 / 11

The Opportunity

Your clients live across borders. Their finances are scattered across countries, currencies, and institutions. No tool gives them a consolidated picture without demanding daily effort — or their bank credentials.

🌍

Fragmented by design

2–4 bank accounts across countries. Income in one currency, rent in another, savings in a third. No single dashboard shows the whole picture.

Existing tools fail them

Cloud aggregators require bank credentials (compliance risk). Budgeting apps demand daily input (low adoption). Spreadsheets work once, then go stale.

What you told us

"The ability to take someone's real financial data and generate personalized output without a heavy compliance burden is exactly what we've been looking for."

03 / 11

Why This Is Possible Now

Three capabilities that didn't exist two years ago converge to make this product viable.

📄

AI reads documents

LLMs can now read a PDF bank statement — any format, any country, scanned or digital — and extract every transaction into structured data. This used to require building a custom parser per bank. Now one model handles them all.

🧠

AI reasons about context

The same model that reads documents can hold a nuanced conversation: "KVBW Wohnen is a German housing association" or "round-number ATM withdrawals usually mean cash payments to household staff." It knows the world well enough to categorize your client's financial life through dialogue.

💻

AI runs on the desktop

Tools like Claude Code run directly on a user's computer, reading local files and executing tasks without uploading data to a cloud service. The AI orchestrates locally — your client's data doesn't need to live in someone else's database.

Your clients already use AI

Many of your clients are already using ChatGPT, Claude, or Gemini for work and personal tasks. The interaction model — asking questions in natural language, uploading documents, iterating on results — is familiar. This product meets them where they already are, applied to a problem they can't solve with a general-purpose chatbot alone.

The window is now. AI-powered financial tools are emerging fast. The cross-border space is underserved by both traditional fintech and generic AI products. There's an opportunity to define what "AI-powered financial clarity" means for expats — with your credibility and domain expertise behind it.

04 / 11

What We've Already Proven

We built a working prototype for a real household — 3 accounts, 2 currencies, 3,800+ transactions, 12 months of data. Here's what we learned.

  • No line-by-line review needed. An AI agent read all the transactions, asked lifestyle questions ("what are these ATM withdrawals for?"), and derived 400+ categorization rules from the conversation.
  • Multi-currency consolidation works. EUR and USD transactions unified with monthly ECB rates. One base currency, one picture.
  • Bundled charges resolved on demand. Amazon orders (195 parsed), Apple subscriptions (63 items), and ATM cash — all unbundled through conversation, not configuration.
  • One self-contained HTML report. Interactive charts, category drilldowns, actionable findings, transparent methodology. Works on phone. Shareable with a passcode.

The core insight: The user described their life. The AI mapped the money. 7 rounds of conversation replaced weeks of manual categorization.

The demo you saw was an anonymized version of this real analysis — same structure, same interactivity, synthetic data.

05 / 11

How It Works

Four steps from bank exports to a shareable, branded report.

Step 1 Client exports data Downloads bank statements (PDFs or CSVs), receipts, invoices. No credentials shared, no API connections. Just files.
Step 2 AI conversation The agent reads the data, asks lifestyle questions, and categorizes everything through a natural conversation. 5–7 rounds.
Step 3 Report generated A single interactive HTML file: spending breakdown, trends, findings, methodology. Branded with your practice identity.
Step 4 Review together Client shares the passcode-gated report with you. Both walk into the planning meeting looking at the same picture.

First report: under 1 hour for single-currency, under 3 hours for multi-currency. Monthly refresh: under 5 minutes — the agent remembers prior rules.

06 / 11

How Client Data Is Handled

Two phases with different privacy profiles. We're transparent about both.

Phase A: Document Extraction — when PDFs need AI reading

Tried first (local)

  • Local PDF text extraction
  • Table detection (tabula)
  • Format-specific parsers
  • CSVs always parsed locally

Fallback (cloud LLM)

  • Complex/scanned PDFs sent to AI for extraction
  • Document content does reach the API
  • User consents per document before sending
  • Happens once per document — never re-sent
Phase B: Categorization Conversation — Agent Bridge protects data

Stays on client's machine

  • Account numbers & IBANs
  • Exact transaction amounts
  • Client name & address
  • Raw CSV/extracted data
  • Generated report & outputs

Sent to AI (anonymized only)

  • Merchant names (public business names)
  • Pattern summaries, amount ranges
  • The lifestyle conversation thread
  • Proposed categorization rules

Every AI interaction is logged locally — both extraction calls (full documents) and conversation calls (anonymized). Reviewable by client, advisor, or compliance counsel at any time.

Anthropic API: No training on data. 30-day retention for safety (or Zero Data Retention for enterprise — content processed but never stored). Exposure decreases over time as local parsers improve for known bank formats.

07 / 11

What This Means for Your Practice

Compliance-light by architecture

  • No client data flows through your systems. The client runs the tool on their own machine. AI processing uses Anthropic's API (no training on data, enterprise zero-retention available).
  • Descriptive, not prescriptive. The report shows patterns and observations. You add the advice in a compliant, documented conversation.
  • No bank credentials. No Plaid, no screen scraping, no OAuth tokens. Zero custody risk.
  • Fully auditable. Every AI call — both document extraction and categorization — is logged locally. Compliance counsel can review exactly what data crossed the network and when.

Fits both your brands

Passport To Wealth

Educational tier. Spend analysis as a coaching tool. Findings framed as observations. Lower price point. Broader audience.

Connected Financial Planning

Full advisory tier. Deeper analysis, forward projections, advisor review mode. Findings framed for planning conversations. Premium positioning.

08 / 11

The Client Experience

A desktop app with a single, always-present input bar. The client talks to the AI the same way they'd talk to you — in plain language.

📁

Drop files anywhere

Bank CSVs, Amazon PDFs, Apple receipts — drag and drop onto any screen. The app routes them automatically.

💬

Natural conversation

"Geraldine is our babysitter, €20/hr in cash." The AI derives categorization rules from lifestyle context, not transaction-level tagging.

📸

Visual feedback

While viewing the report, circle something that looks wrong, type what's off. The AI gets the screenshot, annotation, and data context together.

📊

Interactive report

Charts, drilldowns, findings, methodology. Self-contained HTML — works on phone, desktop, offline. Your brand, your colors.

🔒

Passcode-gated sharing

Client sets a passcode, gets a URL. No login required for you to view it. Open the link, enter the code, review the report.

🔄

Quick refresh

New quarter? Drop in fresh exports. The agent remembers all prior rules — only asks about genuinely new merchants.

09 / 11

Roadmap

Three phases from pilot to scale. We start with your first clients.

Phase 1 — Pilot

First 3–5 Clients

  • Multi-format CSV ingestion (US, EU, OFX)
  • AI-mediated categorization conversation
  • Multi-currency consolidation
  • Branded HTML report generation
  • Privacy-preserving Agent Bridge
  • Compliance-safe observation framing
  • Semi-custom delivery (we assist each client)
Phase 2 — Product

Self-Service Desktop App

  • Guided desktop UI (not CLI)
  • Rule persistence across refreshes
  • Multi-period comparison (YoY, QoQ)
  • Advisor review mode
  • Bank export guides per institution
  • Cross-border insights (FX costs, duplicate services)
Phase 3 — Scale

Agent-Mediated & Multi-Client

  • Automated refresh via client's AI agent
  • Practice dashboard (client roster, report status)
  • Community rule packs by market
  • Anonymized benchmarks across client base
  • Local LLM option (zero network data)
  • Multiple report formats (PDF, slides, data export)
10 / 11

Business Model — Open Questions

There are several ways to bring this to market. The right model depends on how you want to position it relative to your practice and your clients. Let's explore the options.

Option A: Bundled into Connected FP

The report becomes part of your advisory service. Clients get it as part of working with you — it's a differentiator, not a separate product.

Upside: Deepens your value prop. No separate pricing conversation with clients.
Consideration: Cost absorbed into advisory fees. Limited to your client base.

Option B: Offered through both brands

Two tiers matching your two brands. Passport To Wealth gets an education-framed version (spend analysis as coaching). Connected FP gets the full advisory version with planning tools.

Upside: Broader reach. P2W tier can attract leads that convert to Connected FP.
Consideration: Two pricing tiers to manage. Compliance framing differs per brand.

Option C: Standalone product

A new product brand — an AI-powered financial clarity tool for cross-border individuals. You're a co-founder, not just a customer. Sold directly to consumers or through advisor networks.

Upside: Larger market. Product company equity. Your expertise shapes the category.
Consideration: Bigger lift. Needs its own brand, marketing, and support.

These aren't mutually exclusive. You could start with Option A (bundled, pilot with your clients), learn what works, and evolve into B or C as the product matures. The architecture supports all three — branding, compliance framing, and report depth are all configurable.

The bigger question: Is this a tool that makes your practice better — or is this a new kind of product in the cross-border space? LLM-powered financial tools are emerging fast. There may be a window to define what "AI financial clarity" looks like for expats, with your credibility behind it.

11 / 11

Next Steps

Let's start with your clients and build from there.

1

Pilot with 3 clients

Select 3 clients with multi-currency, multi-account situations. We run the full journey together — data to report — and learn what works.

2

Compliance review

Your compliance counsel reviews the report format, findings language, and the Agent Bridge audit log. We establish the "data tool vs. financial advice" boundary together.

3

Build the product

Pilot learnings become the product. Desktop app, self-service onboarding, your branding, your clients. Phase 2 by the end of the year.

Rafael Fernandez Valdes · rafa@rafael.fyi · Berlin