Natural Language Financial Reasoning & Assistant
Ask complex financial questions in plain English. Analyze spending trends, draft budget adjustments, and query your transaction memory with zero hallucinations.
The Problem with Generic AI Chatbots in Finance
- Generic LLMs lack context about your specific accounts, budgets, and past financial decisions.
- Financial advice chatbots often hallucinate numbers or make reckless assumptions.
- Sending raw financial transaction rows to external AI APIs violates personal privacy.
The Moza Way: Grounded Financial Reasoning & Memory
- Moza AI operates from high-level financial signals and structured read models, not raw un-redacted ledgers.
- Financial Memory remembers ongoing goals, life changes, and preferences across sessions.
- Executes real ledger actions (create budget, move money, log bill) with explicit user confirmation.
Key Capabilities
- Natural Language Ledger Actions: Say 'Add $45 for groceries on Chase card' or 'Move $100 from Dining to Travel' to execute actions instantly.
- Interactive Moza Chart Generation: Generates dynamic breakdown charts, line trends, and spending comparisons right inside the conversation.
- Financial Memory Timeline: Remembers past context, upcoming travel plans, and major financial decisions over time.
- Model Context Protocol (MCP): Connect external AI desktop clients (Claude Desktop, Cursor) to your Moza workspace via secure MCP.
Frequently Asked Questions
Is my financial data uploaded or used to train public AI models?
No. Ask Moza uses strict zero-retention API endpoints and privacy-first Layer 2 read models. Your personal financial data is never used to train public models.
How does Ask Moza guarantee 100% mathematical accuracy on spending queries?
Moza AI uses deterministic visualization and retrieval routers. Charts and spending totals are computed directly by the database analytics engine (grounding.check_grounded), eliminating AI hallucinations.
What is Layer 3 Financial Memory and how does it store behavioral patterns?
Financial Memory extracts time-aware observations (e.g. bi-weekly payroll schedules, seasonal utility spikes, recurring subscriptions) to provide contextual answers grounded in your financial history.
Can Ask Moza execute actions like moving money between envelopes or logging bills?
Yes. When you ask Ask Moza to re-allocate funds or record a transaction, it drafts a confirmation proposal card. Nothing is saved until you explicitly click 'Confirm Action'.
How do I connect Claude Desktop or Cursor to Moza using Model Context Protocol (MCP)?
Moza includes a native Model Context Protocol (MCP) server. You can add Moza's FastMCP endpoint to Claude Desktop or Cursor to inspect workspace insights and run natural language queries locally.
How are AI Credits consumed when asking questions or generating forecasts?
Simple data retrieval and chart generation consume 1 AI Credit. Complex 90-day cash flow forecasts or multi-scenario projections consume 5 AI Credits per request.
Can Ask Moza analyze spending across multi-currency accounts?
Yes. Ask Moza automatically consolidates foreign currency accounts (EUR, GBP, JPY) into your workspace base currency using historical exchange rates.
What happens if I start a new conversation in Ask Moza?
Clicking 'New Conversation' resets the active chat session and clears temporary AI draft proposals. It never modifies your underlying transactions, accounts, or budget balances.
Back to All Features or explore Pricing.