If your money lives in five different places — a bank account here, a brokerage there, an ETF you half-remember buying, a mortgage ticking away in the background — you know the pain of trying to answer a simple question: what am I actually worth right now? Aurelio is an open source AI personal finance app that answers that question on your own computer: it tracks your net worth, investments, ETFs, cash and debts locally, and adds an AI financial advisor you can talk to. With over 1,000 GitHub stars roughly two weeks after launch, it is one of the most interesting self-hosted finance tools of the year. Here is what it does, how it works, what it costs to run, and where it honestly falls short.
What is Aurelio? The open source AI personal finance app that keeps your money data at home
Aurelio, built by LosaLosSantos, describes itself in one line: “Open-source personal finance app with an AI financial advisor: track your net worth, investments, ETFs, cash and debts on your own computer.” Technically, it is a Python web app with a modern frontend, shipped under the business-friendly MIT licence. As of 21 October 2026 the repository holds over 1,000 stars and 190 forks — remarkable for a project created on 6 October, about two weeks ago — and it is being updated daily.
The core design decision is the one that matters most: your records stay on your machine, in a local database file (backend/data.db). The AI part is entirely optional and runs through your own OpenRouter key. Nothing about your finances leaves your computer unless you explicitly ask the AI a question. For anyone uneasy about handing bank credentials to cloud aggregators, that architecture is the whole pitch — and it is the same privacy-first instinct behind the local AI tools we track in our AI Tools Directory.
What Aurelio can do
Aurelio covers the full personal-finance loop, not just budgeting:
- Wealth — your banks and brokers, their cash balances, and dated snapshots of what you hold. The “dated” part matters: it keeps history, so you can see how your position evolved.
- Portfolio — every position priced from the market, with its cost basis, unrealised gain and dividends received.
- Look-through — the standout feature: it shows what your funds really hold, broken down by country, sector, company and currency. Two ETFs that look diversified can turn out to be the same ten mega-caps underneath.
- Plans, homes and debts — recurring investments, real assets like property, and the loans financing them.
- Cash flow and goals — income, expenses, savings rate, and the return each goal actually needs to stay on track.
- Ask Aurelio — a chat that answers from your own records, can search the web for context, and proposes every change as a card you confirm or reject.
- The analysis — an analyst and a confidant persona that “know you argue over your portfolio”, with a synthesis that tells you what to look at first.
The project wiki goes further with genuine educational depth: guides on how to calculate net worth, ETF overlap and portfolio X-ray, accumulating vs distributing ETFs, how many months of emergency fund you need, and what an AI financial advisor costs. This is not a thin wrapper — someone who understands the domain built it.
How Aurelio works: setup, usage and the AI part
You need Git, Node.js 22 or later, and uv (the fast Python package manager). Trying it on invented data takes three commands:
git clone https://github.com/LosaLosSantos/aurelio-finance.gitcd aurelio-finance./start.sh --demo (macOS and Linux)
On Windows, the equivalent is .start.ps1 -Demo in PowerShell. The first start installs everything inside the project folder and opens http://localhost:8000 on an invented household, kept in its own database file — so you can explore every screen without touching your real numbers. Run ./start.sh without the flag when you are ready to use your own data.
Two details show engineering care: before any update that changes the database format, the app saves a backup copy next to backend/data.db — your financial history is never one bad migration away from corruption. And the development workflow is standard and healthy: uv run pytest in backend/, npm test in frontend/, with a CI badge on the repo.
The AI advisor is opt-in. You create an OpenRouter account, add credit, generate a key, then copy backend/.env.example to backend/.env and set OPENROUTER_API_KEY. The default model is Claude Opus 5.5 (about $4 per million input tokens, $20 per million output), and you can switch models inside the chat. The author measured real costs in October 2026: a first question that produced a proposal card cost $0.11–0.12, a simple follow-up about $0.02, a full portfolio analysis about $0.64, and each web search roughly three-quarters of a cent. That is honest, verifiable pricing — not “AI magic, cost unknown”.
Why Aurelio is trending now
Three trends are converging on this project. First, finance-data privacy fatigue: the standard way to get a unified money dashboard has been handing your bank credentials to a cloud aggregator. A self-hosted app with a local database is the counter-move, and it resonates with exactly the developers starring repos like this. Second, the AI advisor wave: people want to ask “can I afford this?” or “is this purchase worth it?” against their real numbers, not a generic calculator. Third, the ETF investing boom — the demo animation talks about 1,235 euros left each month, and the wiki’s accumulating-vs-distributing ETF guide is clearly written for European investors. ETF overlap analysis is precisely the feature retail brokers do not give you.
The interaction design deserves credit too. Aurelio never lets the AI change your records directly: it proposes every change as a card you confirm or reject. That is the right UX for money — the agent proposes, the human disposes. Combined with daily commits and a genuinely useful free tier (everything except the AI works with no key at all), the 1,000-star fortnight makes sense.
From our experience: the honest trade-offs
We have spent nine-plus years building data dashboards for businesses, across 300+ projects, and self-hosted finance apps live or die on one thing: data entry. Aurelio’s records are manual, dated snapshots — there is no automatic bank or broker sync in the repo. That is the privacy trade-off, and it is a real one. The tool is superb for the disciplined person who will do a monthly snapshot ritual; most people stop updating after month two and the dashboard slowly becomes fiction. Go in with eyes open: the look-through and goal-tracking features are only as good as the data you feed them.
On cost: the AI is cheap per question but not free, and a weekly full analysis at $0.64 each is about $33 a year — trivial next to a human advisor, but worth knowing before you wire the key in. On advice quality: the AI is an assistant, not a fiduciary. It can be wrong about tax rules, which differ by country, and the wiki guides lean European. Verify anything that moves real money. And on maturity: this project is two weeks old. Expect rough edges and schema migrations — the automatic database backups suggest the author expects them too. That is normal for a young open-source project, not a red flag, but do not treat it as battle-tested infrastructure yet.
One more privacy note: enabling the AI means your questions — phrased against your financial data — travel to OpenRouter and the model provider. Your records stay local, but your queries do not. For most people that is an acceptable trade; for the truly paranoid, the non-AI app remains fully usable. If you want this kind of AI assistance wired into your own business systems instead, our AI integration and automation team builds exactly that.
Getting started with Aurelio
- Install the prerequisites: Git, Node.js 22 or newer, and
uv. - Clone the repo:
git clone https://github.com/LosaLosSantos/aurelio-finance.gitandcd aurelio-finance. - Try the demo first: run
./start.sh --demo(or.start.ps1 -Demoon Windows) and openhttp://localhost:8000. - Explore the invented household: check the net-worth view, the portfolio with dividends, and the fund look-through by country and sector.
- Ask Aurelio something: try “is this purchase worth it?” and watch how it proposes a card you confirm or reject.
- Go live with your own data: run
./start.shwithout the flag and enter your real accounts, debts and goals. - Optional — enable the AI: add an OpenRouter key to
backend/.envasOPENROUTER_API_KEYand pick your model in the chat.
Frequently asked questions
Is Aurelio free to use?
Yes. Aurelio is MIT-licensed open source, so the app itself is free forever, including self-hosting. The only cost is optional: if you enable the AI advisor, you pay OpenRouter for your own usage — roughly $0.02 per follow-up question and about $0.64 for a full portfolio analysis, based on the author’s measured figures.
Where is my financial data stored?
In a SQLite database file (backend/data.db) on your own computer. Nothing is uploaded to any cloud service. The app even backs up the database before updates that change its format. Only the optional AI chat sends data out — your questions go to OpenRouter and the model provider you choose.
Can Aurelio replace a human financial advisor?
No, and it does not claim to. Aurelio is a tracking and analysis tool with an AI assistant: excellent for organising your finances, spotting ETF overlap, and modelling goals. It is not regulated financial advice, and the AI can be wrong — especially on country-specific tax rules. Use it to get organised, then verify decisions that move real money.
Does Aurelio connect to my bank automatically?
Not currently. Records are manual, dated snapshots that you enter and update yourself. That is the deliberate privacy trade-off: no credentials leave your machine, but you do the data entry. It suits disciplined users who will maintain a monthly snapshot habit.
How much does the AI advisor cost to run?
Very little per interaction. Measured in October 2026 on the default Claude Opus 5.5 model: a first question that drew a proposal card cost $0.11–0.12, a simple follow-up about $0.021, confirming a card $0.014–0.049, and a full analysis with lookups about $0.64. Each web search added roughly $0.0075. You can switch to a cheaper model in the chat to cut that further.
Aurelio is the rare open-source finance project that gets the fundamentals right: local-first data, honest AI pricing, and an interaction model that keeps the human in charge of their money. It is young, it needs your data-entry discipline, and it is not a replacement for professional advice — but as an open source AI personal finance app you can run entirely on your own machine, it is the most promising launch we have covered this month. If you would rather have a finance dashboard or AI assistant built for your business instead of your household, talk to us — that is what we do.