Discuss a savings goal with a local model and cited offers
At a glance
Slowth Money combines a private local profile with public Canadian product information to explain savings-goal tradeoffs. This workflow is listed under Money & Bills, Savings Goals, Local AI.
- Agents / tools
- Slowth Money, Ollama, Gemma 3 12B
SourceBuilding a tool for these workflows? See weekly rates.
What they did
1) Use a Mac with Node 22+, pnpm 10+, and enough memory for the chosen local model; the author used an M4 with 24 GB RAM. Clone https://github.com/JulienAvezou/Slowth-Money and run pnpm install followed by pnpm dev. 2) Install Ollama, run ollama pull gemma3:12b and ollama serve, and keep its endpoint at http://127.0.0.1:11434. 3) The app initially opens in demo mode. Choose Settings, then Use my own data, to create a separate encrypted profile. Enter income, savings, debts, account information, and one primary savings goal; leave unknown eligibility or contribution room unknown. 4) Review the deterministic goal projection and public offers with their sources and freshness information. Use Ask Slowth to discuss cash flow, missing inputs, or tradeoffs using the local model. 5) Set the benefit threshold for notifications and inspect the supporting source before acting on an opportunity.
How it went
The author reports using it for a week to track goals and local financial-provider offers, without reporting a measured financial gain.
Worth knowing
This is a proof of concept with Canada as its only implemented market. It does not connect to banks or move money.
