Built for pipelines.
Ready for agents.
RTIO lives wherever your data flows. Connect in-browser via Model Context Protocol (MCP), ingest any structured or unstructured source, drop into n8n workflows, or automate headless pipelines. Node and feature reference
Metadata-only MCP tool calling in your browser.
Most AI tools force you to upload raw datasets to third-party model servers. RTIO provides an In-Browser MCP Client that exposes deterministic DuckDB-WASM tools to your BYOK LLM (OpenAI, Anthropic, Gemini, Groq).
- Zero Raw Egress: Tools only return column types, cardinality, and statistical null rates.
- Deterministic SQL: The model emits executable DuckDB-WASM SQL recipes, never fabricated cell values.
- Cryptographic Audit: Every plan execution appends to an immutable SHA-256 Merkle chain in IndexedDB.
// In-Browser MCP Tool Protocol
[
{
"name":"get_column_schema",
"egress":"metadata_only (0 raw cells)"
},
{
"name":"get_dataset_summary",
"scope":"null_rates, cardinality, bounds"
},
{
"name":"apply_cleaning_step",
"output":"deterministic_sql_recipe"
}
]
Ingest anything. Clean everything.
From structured tabular formats and messy spreadsheets to scanned invoices and cloud buckets, RTIO handles the entire intake spectrum in-browser.
Tabular Structured
Parse high-volume structured files with instant columnar projection pushdown.
Document Extractor
Extract structured data tables from unstructured documents and scanned paperwork.
Cloud Object Storage
Stream datasets directly from enterprise cloud buckets with credential encryption.
REST APIs & Webhooks
Pull live payloads from authenticated endpoints with automatic nested JSON flattening.
Drop into your workflows.
230,000+ engineers use n8n to automate their pipelines. RTIO provides the missing piece: a dedicated Data Quality node that ensures the data flowing between your tools is actually clean.
Native to your terminal.
Run RTIO directly in your CI/CD pipelines, Airflow DAGs, or Docker containers. Same deterministic engine. Same audit trail. Zero UI required.
$pip install rtio-cliCollecting rtio-cli...Successfully installed rtio-cli-1.0.0
$rtio clean sales_data.csv --output clean.parquetProfiling 12,847 rows × 23 columns...Plan: dedup, null_fill, date_std, type_cast✓ Done in 0.31s → clean.parquet (2.1 MB)
Start building robust pipelines.
Free tier includes unlimited local compute. No credit card required.