Table Types

Every Cloudsquid project contains tables. Tables are typed — the type determines what operations are available and how data flows through them.

​

Extraction Tables

An Extraction Table stores uploaded documents alongside the structured data extracted from them. Each row corresponds to one file. When to use: You have PDFs, images, spreadsheets, audio, or video files and want specific fields pulled out into structured columns. Key API operations:

• Upload a file → POST /projects/{name}/tables/{id}/files → returns row_id
• Start an AI run → POST /projects/{name}/tables/{id}/run → returns run_id
• Poll for results → GET /projects/{name}/tables/{id}/run/{run_id}
• Extract synchronously → POST /projects/{name}/tables/{id}/extract

Settings: active_pipeline (flash vs pro), bounding_boxes (source-location highlighting), review_mode (human approval gate).

---

​

Reconcile Tables

A Reconcile Table validates or matches rows of data using an AI agent. Each row is one reconciliation task — the agent compares your input against reference data and returns a structured result. When to use: You’ve extracted invoice line items and want to match them against purchase orders, or you want to validate extracted fields against a known reference dataset. Key API operations:

• Create a task (without running it) → POST /projects/{name}/tables/{id}/tasks → returns task_id
• Run reconciliation synchronously → POST /projects/{name}/tables/{id}/reconcile
• Use the async run pattern with row_id from a task

Input format: AgentJobInput — pass files (references to extraction table rows by UUID) and/or data (arbitrary JSON payload).

---

​

Storage Tables

A Storage Table holds reference data — CSVs or row-by-row JSON inserts. It acts as the lookup source for reconciliation agents. When to use: You need to maintain a table of vendors, product codes, exchange rates, or any reference dataset that reconciliation agents query against. Key API operations:

• Upload or overwrite a CSV → PUT /projects/{name}/tables/{id} (mode: overwrite or append)
• Insert rows as JSON → POST /projects/{name}/tables/{id}/data
• Read rows → GET /projects/{name}/tables/{id}/data

---

​

How they work together

The canonical pipeline uses all three table types in sequence:

1. Storage Table — load your reference data (vendor list, product catalog, etc.)
2. Extraction Table — upload documents and extract structured fields
3. Reconcile Table — pass extracted rows into reconciliation, matched against the storage table

This pattern covers the full lifecycle: ingest → extract → validate.

---