Skip to main content

Neon DB

TalkingSchema connects directly to your Neon serverless PostgreSQL database and imports your live schema as a fully interactive ERD. Neon's branching model makes it an ideal pairing with TalkingSchema: create a Neon branch for schema experimentation, design changes in TalkingSchema with AI assistance, validate the diff, and deploy the migration SQL to your Neon main branch.

Connecting to Neon​

Step 1 — Get your connection string​

In the Neon console:

  1. Select your project and database
  2. In Connection Details, copy the connection string: 
    postgresql://user:password@ep-[endpoint-name].us-east-2.aws.neon.tech/dbname?sslmode=require
  3. Neon requires SSL for all connections — TalkingSchema enables this automatically.

Step 2 — Connect in TalkingSchema​

  1. Click Import → Connect to database
  2. Select Neon
  3. Paste your connection string
  4. Click Connect

TalkingSchema reads schema metadata only. No row data is accessed. Connection credentials are used for this session only and are not stored.

Neon Branching + TalkingSchema Workflow​

Neon's branching feature lets you create instant copy-on-write clones of your database. This integrates naturally with TalkingSchema's schema design workflow:

1. Create a Neon branch (instant, copy-on-write clone of production)
2. Connect TalkingSchema to the branch database
3. Design schema changes with the AI copilot — Plan Mode for visibility
4. Review the diff on the ERD canvas
5. Export migration SQL from TalkingSchema
6. Apply the migration to the branch: validate correctness
7. Apply the same migration to Neon main / production
8. Delete the branch

This workflow gives you a live database to validate migration SQL against before committing to production — at zero storage cost, thanks to Neon's copy-on-write architecture.

Creating a Neon branch for schema design​

# Using Neon CLI
neon branch create --name feature/add-carbon-tracking --project-id [your-project-id]

# Get the connection string for the new branch
neon connection-string --branch feature/add-carbon-tracking

Connect TalkingSchema to this branch connection string, design and validate your changes, then:

# Apply the TalkingSchema-generated migration to production
psql $NEON_MAIN_CONNECTION_STRING -f ./migrations/V012__add_carbon_tracking.sql

# Delete the branch when done
neon branch delete feature/add-carbon-tracking

What Gets Imported from Neon​

Neon uses PostgreSQL, so the import coverage matches the full PostgreSQL import set:

Schema elementImported?
Tables and columns✅ All tables in the connected schema
Column data types✅ All PostgreSQL types including UUID, JSONB, TIMESTAMPTZ, arrays
Primary keys✅ Including composite PKs
Foreign key relationships✅ Including cascade rules
Unique constraints✅
Check constraints✅
PostgreSQL enum types✅
Indexes✅ User-created indexes
Neon-specific extensionspgvector and other extensions appear as column types

Frequently Asked Questions​

Does TalkingSchema support Neon's pgvector extension?​

Yes. If your Neon database uses the vector column type from the pgvector extension, TalkingSchema imports these columns and the AI copilot understands vector-similarity search patterns. Ask: "Add a product_embeddings table with a 1536-dimension vector column for semantic search, indexed with ivfflat."

Does TalkingSchema work with Neon's serverless driver?​

TalkingSchema uses a standard PostgreSQL connection for schema introspection — not the Neon serverless HTTP driver. Use the standard connection string (not the @neondatabase/serverless driver) when connecting TalkingSchema.

What is the connection timeout for Neon's auto-suspend?​

Neon databases may auto-suspend after a period of inactivity. If your connection attempt fails, wait a few seconds for the compute to wake, then retry. TalkingSchema will prompt you to reconnect if the session is interrupted.