Spile

Safely pipeto CSV files

Spile sits between your agents and data, shielding prod data from leaks, edits, and runaway queries.

MCP Clients
Databases
Your app
S
Files
Code Agents
Documents

Features

Why not point agents straight at Snowflake?

Spile acts as a secure database gateway, offering powerful features for AI agent connectivity.

Instant branches & time-travel

Spin up a throw-away branch and query any moment in history—no risk to prod.

Production DB
Spile Branch
AI Agent

Runs in its own sandbox

Queries execute in a lightweight DuckDB container—zero extra load on your warehouse.

AI Agent
Spile container
Your data

Human-in-the-loop approvals

Route prod queries through manual or auto-approval before they ever touch data.

Incoming Query
Spile + Team Review
To Production DB
CSV
Parquet
JSON
Postgresql
Snowflake
S3
Iceberg
DuckDB

How it works

import snowflake.connector
conn = snowflake.connector.connect(
    account='xy12345.us-east-1',
    user='username',
    password='password',
    host='localhostcomputing.com',  # [✓] Use local compute
    warehouse='COMPUTE_WH',
    database='DEMO_DB'
)
cursor = conn.cursor()
cursor.execute("SELECT * FROM users")
Python
Iceberg Metadata
DuckDB Compute
Filesystem Cache
SnowflakeSnowflake

DuckDB runs the query locally

01

Ingest Data

Create Iceberg tables to ingest your data with two way sync with Snowflake.
02

Transform Data

Transform data with SQL or use tools such as dbt and SQLMesh.
03

Analyze Data

Speed up your BI tools and reduce costs by using Spile as a cache layer on top of your Snowflake account.
04

Activate Data

Use Reverse ETL such as Hightouch and Census to load data from Snowflake to your favorite tools.

Latest from our Blog

View all posts