VECTORIO
PROTOR · the engine
ETL here, a warehouse there, a scheduler on top — and nobody can replay last Tuesday.
One engine. The whole pipeline.
Bring data in, clean it, combine it, check it and deliver it — with one tool your whole team understands, on any machine you own.
350+ nodes. One canvas.
Every step you need, ready to drop onto the canvas or write as code — and 170+ of them connect to the systems you already run.
Kafka Subscriber · PostgreSQL Data Reader · S3 Object Reader · Salesforce SOQL Reader · IMAP Subscriber · MQTT Subscriber
PostgreSQL Data Writer · Parquet Publisher · Kafka Publisher · S3 Object Writer · Excel Workbook Writer
Compute Fields · Lookup · Hash Join · Schema Conform · Enrich Data
Window Functions · Rollup Aggregation · Checkpoint · Deduplication · FactOps Throttle · FactOps Debounce
Evidence Collector · Evidence Pack · Evidence Verifier · Column Encryption · Completeness Auditor
ONNX Classifier · Local LLM Processor · Diarizer · Body Pose · Biomarker · LLM Classifier
GDAL Ingest · GDAL Reprojection · AOI Ingest · GeoJSON API Publisher · Delta Composer · Bi-Temporal Diff
Face Blur · RTSP Stream Input · Video Viewer · Audio Reverb · Parametric EQ · Speaker Publisher
Fork · Execution Control Node · Command Control Node · FactOps Gate
What's in the box.
Everything in one place
Loading, cleaning, SQL, live streams and scheduling run in one process.
Canvas and code
Draw a workflow or write it as text — one versioned file.
Connects to what you have
Files, databases, streams, cloud storage, geodata and sensors — 170+ connectors, 350+ nodes.
Runs on your machines
Windows, Linux and macOS, on x64 and ARM64 — offline if you need it.
Picks up where it stopped
Checkpoints and crash recovery. Rerun any job and get the same result.
Proof built in
Every run leaves a tamper-evident record, and results can be signed.
Every result explains itself
Each run leaves a hash-chained record; outputs can be signed and verified offline, without us.
Workflows at real scale
Up to 10,000 nodes in one real-time workflow — the Beethoven demo runs at 48 kHz with no buffer underruns.
Runs where the data is
Notebook, server, edge box, Raspberry Pi-class board (beta) or a cluster (beta). Offline if required.
Measured on one Apple M3 Max-class notebook, 16 cores, 64 GB, CPU only; 1 cold + 5 warm runs; logical decoded GB/s. Method and raw table →
Bring us your hardest data problem. Let's solve it together.
A short pilot on your data and your hardware. We agree on what success looks like before we start.