Personal finance dashboard with LLM

Pain

Financial decisions need numbers nobody keeps in their head: cashflow forecast, tax levers on profit spikes, long-term investment effects. Classically these answers wait for the quarterly meeting with the tax advisor, too late for tactical calls.

Solution

• ◆Consolidate existing financial data (accounting, banking, contracts) into a single base.
• ◆Visualize historical and forecasted cashflows in a dashboard with clear views.
• ◆Connect an LLM to the data: it can run forecasts and surface levers (example logic: on a profit spike, mention §7g IAB reserve or special depreciation).
• ◆Deliver the answers on demand instead of on a quarterly cycle.

How it works

1. 01
   Data base

   Own finance data is consolidated into a single table/database.
2. 02
   Visualization

   Power BI renders cashflow, contribution margin, return on investment.
3. 03
   LLM connection

   Claude receives the relevant data slices as context and answers concrete questions.
4. 04
   Tax-lever hints

   On distinctive patterns (e.g. profit spike), the LLM lists possible levers with paragraph references, not advice, but preparation for the tax advisor.

Outcome

The forecast is available any time, not only at the quarterly meeting. Concrete amounts stay with the tax advisor; the dashboard supplies the questions for that conversation.

Why this stack

Claude API

Reasoning on structured data slices with transparent justifications. Our own finance data, our own call; for clients an open-source model on our own GPU or an approved provider.

Python

Data preparation and forecasting logic in a single stack.

Power BI

Familiar visualization with drill-down, shared with the tax advisor.