A local analytics workbench for large data.
Import a file of your own or pick a public dataset, then pivot, filter and sort millions of rows live. The database runs in this browser tab: no server, no account, and your files are never uploaded.
Use a laptop or desktop browser. A large dataset downloads once, then opens from this device.
| Airline | Flights | Avg delay | Max delay |
|---|---|---|---|
| Total | 5,819,079 | 4.4 | 1,971 |
| WN | 1,261,855 | 4.4 | 659 |
| DL | 875,881 | 0.2 | 1,274 |
| AA | 725,984 | 3.5 | 1,971 |
| OO | 588,353 | 5.8 | 1,372 |
| EV | 571,977 | 6.6 | 1,223 |
| UA | 515,723 | 5.4 | 1,294 |
| MQ | 294,632 | 6.5 | 1,528 |
| B6 | 267,048 | 6.7 | 1,002 |
| US | 198,715 | 3.7 | 750 |
| AS | 172,521 | −1.0 | 950 |
| NK | 117,379 | 14.5 | 833 |
| F9 | 90,836 | 12.5 | 1,101 |
| HA | 76,272 | 2.0 | 1,467 |
| VX | 61,903 | 4.7 | 651 |
Your data
Press Import, or drop a file on the window. CSV, TSV and text, JSON and JSON Lines (JSONL, NDJSON), and Arrow or Feather load into the database; Parquet of any size is queried in place, read from your disk as needed. All of them pivot, filter and sort. Anyi reads each column’s type and how many values it holds, so only columns that make sensible groups are offered for rows and columns.
Public datasets, millions of rows
Or start from one of sixteen public datasets. Each opens as every row, or as a pivot ready to change, with the question it answers and a next step to try. The table draws only the rows in view, so scrolling stays smooth at any size.
- UK house sales, 1995–2023 · HM Land Registry28.3M sales
- Medicare Part D prescribing, 2023 · CMS26.8M rows
- US flight delays, 2015 · US DOT5.8M flights
- Consumer complaints, 2024 · CFPB2.7M complaints
- NYC street trees, 2015 census · NYC Parks683,788 trees
- Online Retail, 2010–2011 · UCI541,909 lines
- Mortgage lending, NY, NJ, CT, 2017 · HMDA386,100 loans
- Federal IT contracts, FY2025 · USAspending266,830 rows
- MoMA collection · Museum of Modern Art160,982 artworks
- Global power plants · WRI34,936 plants
- Company financials from 10-K filings · SEC22,832 facts
- NYC film permits, 2023–2026 · NYC Open Data20,253 permits
- Capital Bikeshare, hourly, 2011–2012 · UCI17,379 hours
- Employment and wages by county, NY, NJ, CT · BLS QCEW16,573 cells
- Earthquakes, last 30 days · USGS9,827 events
- Superstore orders · sample9,994 lines
Group, filter and sort where the data is
DuckDB, compiled to WebAssembly, runs in the page and answers each change. Group rows into a tree, split measures into columns, filter with conditions joined by All or Any, and sort by several keys in order. A column that cannot group, such as a measure or an ID that differs on nearly every row, is listed under Not available with the reason.
Side by side
Each dataset or file opens in a tab of its own. Duplicate an analysis to change a copy apart, and drag a tab to a side to compare two at once. Save an analysis as a view to open later, and export the table as it shows, as CSV. Reloading brings every tab back.
Everything runs in this tab
Anyi has no server and no account. The page loads DuckDB, compiled to WebAssembly, and every query runs on your computer. Datasets come straight from their public hosts and are kept on this device, so the next visit opens them without downloading again. A file you import is read in this browser and never uploaded.