Financial planning
Trellis helps you explore financial decisions through an embedded AI agent. Describe a savings goal, import a bank statement, or ask what buying a home would change. The agent prepares proposed updates to the financial model so you can compare the results. This agent workflow is in active development.
Compare scenarios
Start with your current finances, duplicate them into a named scenario, change an assumption, and compare cash flow, net worth, investments, and goals. Guided scenario builders cover decisions such as a mortgage, a salary change, lower spending, or increased investing.
Projections are recalculated when assumptions change, so each scenario reflects the wider effect of a decision.
Review changes
The agent proposes changes; Trellis checks them and shows what would change before you approve. The app then applies the approved actions to the model. Bank-statement imports follow the same process, with proposed transactions available for review before they are added.
Financial records are stored locally. AI requests use an external model provider and can include financial context needed to handle the request.
How it works
Trellis is an Electron desktop app with a React interface and a local SQLite database. Its TypeScript calculation engine is separate from storage and the interface, making cash-flow, investment, mortgage, and goal calculations directly testable. Projections are derived from assumptions rather than saved as fixed results.
The agent works through the app's validated actions instead of writing directly to the database. This connects conversational requests to the same financial model used by the rest of the app.
Current state
The financial-planning application is working and used locally. The embedded agent and statement-import workflow are in active development and have not been released. Trellis is not distributed publicly today.