- Published on
Show HN: Pulse Local, a Mac budgeting app whose AI models run on your machine
- Authors
- Name
- DeepHorus
Pulse Local, a Mac budgeting app, has launched on Hacker News' Show HN. Its AI chat answers questions from the user's own transactions, with the models running on the same machine. Nothing is sent to an online AI service, because there is no endpoint to send to.
Most local-first finance apps keep the ledger on disk but send the question to a hosted model. A privacy policy may promise not to train on that data, yet the query still leaves the machine. Pulse Local keeps both in place.
The product is a full budgeting app rather than a privacy demonstration. It covers plans and targets, rules, rollover, splits, cash flow and forecast. Its money maths was checked against six months of a real ledger and matched to the cent.
Local mode is global. The currency picker lists 75 countries, every currency is accepted, and no exchange rates are applied; each ledger holds one currency. Figures are written the local way, such as 1.234,56 € in Berlin or ¥1,311 in Tokyo. Bank files import in any language, but the chat answers only in English.
Transparency is built in. Labels above a spending figure show which transactions were counted, and users can change them. When the model's two readings disagree, it says it is unsure rather than guessing, and AI answers are labelled.
The team describes itself as a small group of machine-learning practitioners and product people. Pricing is a one-time payment, not a subscription, and every feature is available free; bank sync is the only item with a running cost.
The developers tune small open models to answer money questions locally, benchmark them openly against cloud models, and publish the weights on Hugging Face. Next are Windows, bank sync for more people, and the answers the model still gets wrong.
