Quick Start

Upload and query data from the terminal, a Python notebook, or inside an AI agent. Follow the steps below to authenticate and run your first query.

CLI

Python SDK

Fast path: create, load, and query

hotdata databases create \
  --name airbnb \
  --catalog airbnb \
  --table listings

hotdata databases load \
  --catalog airbnb \
  --table listings \
  --url https://hotdata.dev/data/sf-airbnb-listings.parquet

hotdata query \
  "SELECT COUNT(id) AS total_rows FROM airbnb.public.listings"

1) Install the CLI

Install walkthrough (YouTube):

brew install hotdata-dev/tap/cli

Verify the installation:

hotdata --help

2) Authenticate

Authenticate via browser:

hotdata auth login

A browser window will open for you to sign in and authorize the CLI. Verify you're logged in:

hotdata auth status

3) Instant databases

Instant databases are Hotdata-owned catalogs you populate with parquet files. Create them on demand, load data, query immediately, and delete when done.

Instant database tutorial (YouTube):

Create a database and declare the tables you plan to load:

hotdata databases create \
  --name mydb \
  --catalog mydb \
  --table orders \
  --table customers

Load a parquet file from a local path or URL:

# From a local file
hotdata databases load \
  --catalog mydb \
  --table orders \
  --file orders.parquet

# From a URL
hotdata databases load \
  --catalog mydb \
  --table orders \
  --url https://hotdata.dev/data/sf-airbnb-listings.parquet

Query the loaded table — managed tables are addressed as <catalog>.<schema>.<table>, where <catalog> is the alias you set with --catalog:

hotdata query \
  "SELECT * FROM mydb.public.orders LIMIT 10"

List databases and their tables:

hotdata databases list
hotdata databases tables mydb

Delete a table or the whole database when you're done:

hotdata databases tables remove orders --database mydb
hotdata databases remove mydb

4) Query your data

Basic query

hotdata query "SELECT id FROM mydb.public.orders LIMIT 5"

The database is resolved automatically from the catalog-qualified table name (mydb.public.orders), so no extra flag is needed. Use -o table|json|csv to change the output format, or --database <id> to target a specific instant database by its id.

Analytical functions

Window functions for rankings, running totals, and row comparisons:

hotdata query "
  SELECT id, amount,
    sum(amount) OVER (
      ORDER BY id
      ROWS BETWEEN UNBOUNDED PRECEDING
        AND CURRENT ROW
    ) AS running_total
  FROM mydb.public.orders
  LIMIT 10
"

hotdata query "
  SELECT date, symbol, price,
    lag(price) OVER (
      PARTITION BY symbol
      ORDER BY date
    ) AS prev_price
  FROM mydb.public.stock_prices
"

Full-text search

Create a full-text index on the text column, then search it by name — the search type is inferred from the index:

hotdata search create articles_body \
  --type text \
  --from mydb.public.articles \
  --column body

hotdata search "machine learning" \
  --index articles_body \
  --select id,title,body \
  --limit 10

Vector search

Create a vector index — --provider auto-embeds the text column server-side. Then search it by name; the server resolves the embedding model from the index metadata:

Vector search demo (YouTube):

hotdata search create documents_body \
  --type vector \
  --from mydb.public.documents \
  --column body \
  --provider <embedding_provider_id>

hotdata search "documents about machine learning" \
  --index documents_body \
  --limit 10

An auto-embedding vector index has to be the only index on its table, so this example indexes documents rather than adding a second index to the articles table used above. If you try both on one table, whichever you create second is rejected. See vector indexes.

Equivalent SQL when you already have a query vector:

hotdata query "
  SELECT id, title,
    l2_distance(embedding, ARRAY[0.1, -0.2, 0.5]) AS dist
  FROM mydb.public.documents
  ORDER BY dist ASC LIMIT 10
"

For cosine similarity, use cosine_distance; for inner product, use negative_dot_product.