Your private data agent
Runs on your machine. Queries everything. Sends nothing to the cloud. Connect S3, databases, local files, and remote sources — then ask questions in plain English. Your data never leaves.
Free, open source (AGPL-3.0) · unlimited sources · runs fully offline · macOS, Windows, Linux
Install on more than one machine? Pair them with a workspace key for cross-machine search and plain-English queries across all of them.
Install
One line. Then it's yours.
Native installers for macOS, Windows, and Linux. No account, no login, no “set up your workspace” first. Run the install and start browsing your storage.
$ brew install seryai/tap/sery-linkor universal installer: curl -fsSL https://sery.ai/install.sh | sh
One app · 9 storage protocols · 4 S3-compatible presets
Connect every storage
9 protocols and 4 S3-compatible presets — Local, HTTPS, S3 (+ B2, Wasabi, R2, GCS), Drive, SFTP, WebDAV, Dropbox, Azure, OneDrive. One app replaces the handful of cloud-console + file-browser tabs you have open right now.
Preview, profile, query — without downloading
Sery reads Parquet footers over the wire — schemas in under 2s on multi-GB files. CSV / TSV / Excel preview streams just enough. Auto-profile (null %, unique values, distributions) computed locally. Convert any tabular file to Parquet in place.
One network across machines
Pair every machine you've installed Sery Link on with a workspace key ($19/mo). Query the whole network in plain English from one place — SQL is dispatched to each machine; only result rows travel back. Raw files stay where they live.
Type a column name. Understand the data behind it.
Sery matches against filenames, column names, and extracted document content across every folder and every remote source you have added. Open a result to see the schema, sample rows, and column statistics. Connect to sery.ai to ask questions across all of it in plain English.
Why this matters
- ✓OS-level file search doesn't know your data: it searches filenames and maybe content — never column headers, never across cloud storage.
- ✓No SQL, no DataFrames: click a result to see schema, sample rows, and column stats (null %, unique, min/max/avg) — all computed for you.
- ✓Plain-English queries when you need them: subscribe to sery.ai to ask across every file Sery knows about — your data network, one question at a time.
- ✓Spans machines + remote sources: a workspace key connects multiple laptops; HTTPS URLs and S3 buckets appear in the same index.
“I used to keep a note saying which spreadsheet had which column. Now I just type the column name.”
— Early user
You have data. You can't get answers from it.
Finding the right file is half the problem; making sense of what's inside it is the other half. The tools that find your files don't understand your data. The tools that understand your data want you to upload it first.
Generic file search
- —Finds files by name; can't tell you what columns they have or what the values look like
- —Doesn't compute anything — null rates, distributions, joins are all on you
- —One disk per search; no way to span other machines or remote sources
Cloud BI tools
- —Upload everything first. Your CSVs become a warehouse problem.
- —Monthly billing on row counts; seats per teammate
- —Privacy headaches — your files live with a vendor now
Sery: the data layer that runs on your laptop
Local-first SQL indexing. Column-aware search. Automatic per-file stats. AI queries across everything when you want them. CSV, Parquet, Excel, DOCX, HTTPS URLs, S3 buckets — all indexed on your own hardware. Nothing uploaded. Nothing per-seat.
Fast, private, and unobtrusive
Sery Link runs an embedded SQL engine on your laptop to index and query your data locally. No upload step. No server between you and your files.
Search in milliseconds
The scan cache is a persistent local SQL index. Every search hits memory, not disk. Even thousands of files feel instant.
Nothing leaves your laptop
Works fully offline with no account. Remote sources (S3, HTTPS URLs) still fetch directly from your machine — we never proxy your data.
Lives in your menu bar
Small native app. Scans watched folders in the background, wakes up when you need it. macOS, Windows, Linux.
How it works
Sery Link walks your watched folders, extracts schema + sample rows via an embedded SQL engine, and caches the result in a local index. Searches hit the cache; previews run on your laptop; remote sources are fetched directly with credentials stored in your OS Keychain. Nothing ever lands on our servers unless you explicitly connect for AI queries.
sery.link
Your data stays put. Your reach is unlimited.
Enable a public portal on any Sery Link machine and anyone can query your live data — without you uploading a thing. Discover and query thousands of live datasets at sery.link.
Publish on sery.link
Flip one toggle in Sery Link. Your machine gets a unique sery.link/<hash> URL. Consumers can browse your schema and run SQL against your live data — your files never move.
Earn tokens as a publisher
Every query against your portal earns you 80% of the token cost. Tokens cash out to real money via Stripe. The more useful your data, the more you earn — no middleman taking a cut.
Query with trust signals
Every portal shows success rate, p95 latency, quality labels, and verified reviews. Know what you're querying before you spend a single token.
Built for people drowning in spreadsheets
Ask your data from your phone
Hold the mic button and ask in plain English. Sery Snap reaches every machine in your workspace and returns the answer — no typing, no SQL, no laptop open.
- Hold-to-record voice input — speak your question, get a plain-English answer
- Sees every machine in your workspace — query across all of them at once
- Full conversation history — revisit any answer, any time
Stop downloading files just to look at them.
Free, open source, runs on your machine. Browse every cloud storage you have, preview tables and Parquet without downloading, run SQL on remote bytes. Add Sery to a second machine? Plus ($19/mo) connects them and adds AI chat across everything. 7-day free trial, no card.