Knowledge Tables

Purpose

Knowledge tables are shared CSV datasets your organization uploads so AI agents can look up structured facts (prices, SKUs, service lists, FAQs in tabular form) during a reply — without stuffing everything into the system prompt.

What you can do

• Create and manage tables under Knowledge tables.
• Upload or edit CSV data (practical limits: size, row, and column caps enforced by the platform).
• Attach one or more tables to an agent on the agent Knowledge step.
• Let the agent use the lookup_knowledge_table tool when a question needs a row-level answer.

Quick start

1. Go to Knowledge tables and create a table.
2. Load CSV columns that match how your team already describes offers or policies.
3. Open an AI agent → Knowledge → link the table.
4. Publish the agent and test a question that should hit a lookup.

Best practices

• Prefer short, dense tables over long prose dumps.
• Keep column names clear (sku, price_mxn, hours) so lookups stay reliable.
• Update the table when offers change — agents read live linked sources.

Related

• AI Agents