What is Loam

Learn how Loam connects your business data to your AI so it can answer from real numbers instead of guessing.

The problem

You ask your AI about last quarter's revenue. It says $4.2M. The real number is $3.87M. Your AI made it up because it doesn't have your actual data.

This is called hallucination. The AI sounds confident, but it's guessing. It happens because your AI can't see your actual data: your transactions, your customer records, or your spreadsheets.

How Loam fixes it

Loam holds your own data and gives your AI access to the real numbers.

1. Upload your files. Add CSV, TSV, JSON, Excel (.xlsx), and Parquet files. Loam brings the tables together so your AI can query your real data.
2. Work from real rows and columns. Loam runs queries against the data you uploaded, so the numbers come from your tables instead of a guess.
3. Connect your AI. Use whatever AI you already have: ChatGPT, Claude, or your own tools. Connect it to Loam and it works with your real business data.

How your data is organized

Loam uses one workspace-rooted hierarchy for organization and SQL:

• Datasets keep related data together and can contain smaller datasets when another level of organization is useful.
• Tables hold your actual data. Each table has rows and columns, like a spreadsheet. When you add a file as a new table, an orders export can become an orders table.
• Views present saved query results alongside tables.

The same dataset path you see in the sidebar is the full SQL path. In SQL, each dataset in that path is called a namespace. For example, a charges table inside commerce.stripe.production is queried as commerce.stripe.production.charges.

What happens next

The fastest way to get started is to bring your own files. Upload a CSV, TSV, JSON, Excel (.xlsx), or Parquet file. When processing finishes, an AI you have connected to Loam can query the data it has permission to see. Follow your first upload for a step-by-step guide.