The DuckHouse blog

The Pocket Warehouse

A field guide to owning your data, in ten parts: from numbers scattered across a dozen tools to answers you can ask for from your phone.

Start at part one RSS

Part 08 · The Pocket Warehouse

Your warehouse.
In your pocket.

Latest · Part 8 · October 16, 2026

Your warehouse, in your pocket

Connect your warehouse to an AI assistant once, then ask business questions from your phone in plain English: the setup, the questions that work, the limits.

Read part 8

All ten parts

Read in order, the series follows one small business from scattered numbers to a warehouse it can question from anywhere. Each part also stands on its own. A new part comes out every week or two, on the dates below.

01What a data warehouse is, and why you might finally need oneA plain-English guide to data warehouses: what they are, how they differ from the database behind your app, and the signs a business is ready for one.October 1
02DuckDB, and why small data got fastWhat DuckDB is, why it runs analytical queries so quickly on a single machine, and how to start asking questions of CSV and Parquet files with it in minutes.October 2
03Inside the engine: how a columnar database reads so littleRows versus columns, compression, vectorized execution and zone maps: how an analytical database like DuckDB answers a question by reading as little as it can.October 5
04Getting your data in: syncs, APIs and CSVs, without a pipeline teamETL or ELT, full or incremental syncs, the updated_at trap, deletes, rate limits and CSV drops: a practical guide to getting business data into a warehouse.October 7
05Shape it once: modeling a warehouse that people and AI can readRaw, clean and mart layers, facts and dimensions, names that explain themselves, and the views that turn synced tables into answers people and AI can trust.October 9
06The SQL behind every business questionRevenue by week, year over year, cohorts, retention, churn and margin: a DuckDB SQL cookbook for the questions every business asks, with examples you can run.October 12
07MCP, explained: how AI assistants reach your dataThe Model Context Protocol in plain English: hosts, servers and tools, how an assistant turns a question into SQL, and the safeguards that keep your data yours.October 14
08Your warehouse, in your pocketConnect your warehouse to an AI assistant once, then ask business questions from your phone in plain English: the setup, the questions that work, the limits.October 16
09DuckLake, explained: a lakehouse you can hold in your headOut October 27
10Build a data warehouse in an afternoonOut November 10

Launch your first warehouse.

No contracts and no minimums. Pick a size and a region, attach from your terminal, and pause it whenever you like.

Get started Read the docs