Multi-source obligation monitoring
Watches Outlook and email, Open Finance accounts and transactions, and scheduled events as inputs to one financial state.
Finance Concierge
A personal finance automation system that watches obligations and rules across Outlook, Open Finance, and scheduled events, then executes explainable actions through MCP interfaces.
Role & scope: Sole engineer: built the obligation monitoring, Open Finance ingestion, scheduled rules, persistence, automated actions, MCP workflows, and interfaces.
Core capabilities
The work behind the concierge.
Rules and automated actions
Turns detected obligations into explainable tasks, reminders, updates, forecasts, and other bounded actions.
MCP interface layer
Makes the same financial system available through Telegram, Claude, and Codex, without making any one interface the product core.
Read this first
The system at a glance.
- What the product does
- A personal finance system that observes obligations across multiple sources and organizes the resulting tasks, actions, and forecasts.
- Who it serves
- People who want recurring financial obligations monitored and handled through dependable automated rules.
- What makes it different
- It separates the financial system from its interfaces: sources feed obligations, rules trigger actions, and Telegram, Claude, or Codex can expose the state.
- My contribution
- Sole engineer: built the obligation monitoring, Open Finance ingestion, scheduled rules, persistence, automated actions, MCP workflows, and interfaces.
- Proof point
- The complete pipeline can be retried without duplicating a bill, task, or message.
Capability flow
How the system works.
Each step connects a user action to the system capability behind it.
-
Observe obligations across sources
Outlook and email, Open Finance accounts and transactions, and scheduled events feed the same structured financial state.
-
Normalize obligations and rules
Provider identities, deduplication keys, and explicit rule state make repeated observation safe to replay.
-
Trigger automated actions
Deterministic rules control reminders, updates, forecasts, and when bounded AI reasoning or MCP tools are useful.
-
Persist organized financial state
Obligations and actions remain explicit, with source, status, amount, and completion state attached.
-
Expose the system through interfaces
Telegram, Claude, and Codex provide ways to visualize and interact with the same underlying financial state.
Engineering judgment
Decisions that shaped the system.
- Constraint
- Polling, retries, and process restarts can deliver the same financial signal more than once.
- Decision
- Use provider identities, unique deduplication keys, and processed/notified markers instead of relying on one transaction.
- Why it mattered
- The complete pipeline can be retried without duplicating a bill, task, or message.
- Trade-off
- Every stage must expose stable identity and completion state.
- Constraint
- AI interpretation is useful, but money-related automation must remain explainable and controllable.
- Decision
- Keep deterministic rules around bounded agent decisions and attach the source and rule to each outcome.
- Why it mattered
- A user can understand why an obligation exists and correct it through the same interface.
- Trade-off
- The agent has less freedom than an open-ended assistant, by design.
Technical evidence
Skills backed by this system.
Open any skill to see the other products and professional work connected to it.
- AI Agents Uses bounded reasoning to interpret financial signals inside an auditable task workflow.
- MCP Uses the Model Context Protocol to expose bounded financial actions and state to Claude and Codex interfaces.
- Deterministic Automation Rules, stable identities, and completion markers keep repeated processing predictable.
- TypeScript Carries typed message, rule, obligation, and notification contracts through the complete automation pipeline.
- Event-driven Systems Moves external message events through ingestion, classification, task creation, and notification stages.
- Node.js Runs the HTTP service, scheduler, polling pipeline, and Telegram bot in one controlled process.
- Hono Provides the operational API and guarded refresh endpoints around the financial pipeline.
- REST APIs Exposes a lightweight API for tasks, reminders, rules, providers, events, and operational controls.
- Testing Uses unit, journey, API, provider, and end-to-end tests to verify the financial pipeline and Telegram workflows.
- Zod Validates configuration, rule definitions, API bodies, provider data, and MCP tool inputs at system boundaries.
- node-cron Schedules Outlook polling, summaries, escalations, backups, and other recurring financial jobs.
- MCP Server Development Provides MCP tools for querying and managing financial state alongside the Telegram and REST interfaces.
- Prisma Models sources, obligations, tasks, deduplication keys, and notification state in the persistence layer.
- Vercel Provides the deployment environment for the financial assistant's web-facing application surface.
- Microsoft Graph Uses delta polling to ingest incremental changes from multiple Outlook accounts efficiently.
- Telegram API Turns stored financial state into reminders, confirmations, reports, forecasts, and interactive commands.
Technology stack
- Node.js
- TypeScript
- Hono
- REST APIs
- Testing
- Zod
- node-cron
- MCP Server Development
- Prisma
- Vercel
- Microsoft Graph
- Telegram
- MCP