Schema Management on Autopilot

Stop fearing upstream API changes. Forge acts as a firewall for your data warehouse, detecting schema drift, versioning changes, and alerting you instantly.

Start Free Trial See It in Action

A Living History of Your Data

✓

v1.0 Created Oct 12, 09:00 AM

Initial Schema Inferred

Forge scanned 5,000 records from stripe_charges and established the baseline schema.

⚠️

Drift Detected Oct 14, 02:15 PM

New Fields Found in Webhook

Upstream API added metadata fields. Forge automatically captured them.

"id": "ch_123", "amount": 2000, + "metadata": { + "campaign_id": "black_friday_2024", + "source": "mobile_app" + }

🔄

v1.1 Applied Oct 14, 02:16 PM

Schema Evolved Automatically

Forge executed ALTER TABLE to add new columns. No data was lost.

ALTER TABLE stripe_charges ADD COLUMN metadata__campaign_id STRING;
ALTER TABLE stripe_charges ADD COLUMN metadata__source STRING;

Stay in the Loop, Not in the Weeds

You don't need to approve every change, but you should know about them. Forge integrates with Slack, Teams, and Email to keep your team aware.

Forge APP Today at 2:15 PM

Schema Change Detected: stripe_charges
We detected 2 new fields in the latest batch. The schema has been updated to v1.1.

New Columns:
• metadata__campaign_id (STRING)
• metadata__source (STRING)

Intelligent Fault Tolerance

Bad data happens. An upstream API might send a string where you expect an object. Forge skips the bad rows and alerts you immediately, while continuing to process the good ones. Your pipeline never stalls.

Incoming JSON Batch

{ "id": 1, "user": { "name": "Alice" } }
{ "id": 2, "user": "INVALID_STRING" }
{ "id": 3, "user": { "name": "Bob" } }

→

✅ Valid Rows Processed

Rows 1 and 3 are normalized and loaded into BigQuery tables immediately.

🛡️ Bad Rows Skipped

Row 2 is skipped to protect the schema, and you receive an instant alert with the error details.

📜

Full Version History

Time-travel through your schema. See exactly what your data looked like last month vs. today.

🛑

Breaking Change Protection

Forge blocks type mismatches (e.g., String → Integer) and quarantines bad rows so your pipeline never crashes.

🔍

Data Lineage

Trace every column back to its source JSON path. Auto-generated documentation for your data catalog.