Numera is for people who've outgrown Excel.
A browser-based data workbook you can open and use immediately. It is free to try, requires no sign up, keeps your data on your machine, and still works offline. It keeps the spreadsheet mental model that power users already understand and removes the parts that break once the model becomes serious.
- Free to use
- No sign up required
- Private by default
- Works offline
Think of it as a spreadsheet that knows it has rows and columns, remembers what they mean, and keeps formula logic attached to labels instead of cell addresses.
No install
Runs in the browser with Python client-side.
No server
No sync button, no cloud roundtrip, no backend dependency.
Local data
Your workbook stays on your machine, even offline.
Formula preview
profit = revenue - costs
total = sum(Q1 .. Q4)
yoy = revenue / revenue[PREV] - 1Live app
Open Numera and start working now.
No sign up. No setup. No waiting for an account. Open the app, load data, and start modeling immediately. Your data stays private on your machine, and the workbook still works offline.
Free
Start using the app without a paywall blocking the first step.
No sign up
No account creation, no email capture, no activation flow.
Private
Your data stays local instead of being shipped to someone else's server.
Offline
Keep working when you have no connection or do not want one.
What is Numera?
Built for Excel power users who already know exactly where Excel breaks.
You do not need to know Python. Numera is for the person who builds real financial and analytical models, has been burned by spreadsheets, and wants something better without becoming a programmer first.
A formula silently broke because someone inserted a row above it.
A total looked right until you realized one cell was hard-coded.
A date column became a number, or a number became text, for no reason.
The model turned into fourteen versions of the same workbook.
One edit in the wrong sheet corrupted the whole thing.
The core idea
Labels, not addresses.
In Excel, formulas point at grid positions. Insert a row or column and the formula can still calculate while now referring to the wrong thing. Numera references meaning instead. It knows what the formula is trying to compute, not just where the cells happened to be.
Excel
=VLOOKUP(A2, B:F, 3, FALSE)
=SUM(B2:E2)
The 3 means the third column in the range. Insert a
column and you get different data with no warning. Add Q5 and the
sum does not expand itself.
Numera
profit = revenue - costs
total = sum(Q1 .. Q4)Reorder columns, insert rows, rename sheets. The formula still works because it points at labeled data, not fragile addresses.
Formulas are visible
In Numera, every formula lives in a dedicated formula surface. You can see them all at once instead of clicking cell by cell to discover hidden logic.
Formulas apply once
Define a formula once and it applies across the relevant row or column automatically. No copy-paste chains down 200 rows.
Examples
profit = revenue - costs
total = sum(Q1 .. Q4)
yoy = revenue / revenue[PREV] - 1Data types that do not lie
Numbers stay numbers. Dates stay dates.
- Every column has a declared type: number, text, date, or boolean.
- You define it explicitly. Numera does not silently reinterpret it.
- Bad input becomes a visible data issue instead of a hidden model bug.
Time series is first-class
Built on pandas, not spreadsheet workarounds.
- Resample daily data to monthly or quarterly with the right semantics.
- Use rolling windows for moving averages and trailing sums.
- Shift and lag periods for year-over-year and prior-period comparisons.
- Dates sort like dates instead of pretending to be text.
The workflow
Five steps, without collapsing everything into one spreadsheet soup.
Step 1
Add
Load data, create DataFrames, and write Python if you want to.
Step 2
Check
Inspect declared column types and catch bad inputs before formulas depend on them.
Step 3
Calculate
Define formulas once and apply them across the full row or column.
Step 4
Graph
Visualize the parts of the model that matter without leaving the workbook.
Step 5
Present
Clean up the view for sharing after the model is already correct.
Python when you want it
The formula language handles the common case. Python remains first-class.
If you know Python, you can manipulate DataFrames directly, use pandas APIs, install packages, and treat the workbook as a serious data tool. Python is the foundation, not a bolt-on.
df_sales["price"] = 99The grid updates, formulas recompute, and the UI reflects the new state without a sync step.
Everything runs in your browser
No account. No cloud sync. No hidden server.
Numera runs Python client-side via Pyodide. Your data stays on your machine. You can work offline. The browser is the runtime, not a thin client talking to someone else's backend.
Numera uses privacy-friendly aggregate GoatCounter analytics to understand basic site usage. It does not use analytics cookies, store IP addresses, or track workbook contents.
Why that matters
For serious workbook users, privacy and control are not decorative features. They shape whether the tool can be trusted with real data.
What Numera is not
It solves the specific parts of spreadsheet modeling that are broken.
It does not try to replace every tool around a model. It is aimed at the person who already knows how they think and wants the workbook itself to stop fighting them.
It is not a replacement for SQL or a database.
It is not a BI tool with dashboard builders, scheduled reports, or warehouse syncing.
It is not a collaborative editor yet.
It is not Excel. The model is labels instead of addresses, and formulas live in a dedicated surface instead of cells.
Try Numera now, without creating an account first.
Free to use. No sign up needed. Your data stays private on your machine and the app works offline. If that matters to how you work, the fastest way to understand Numera is to open it.