Connecting Data Sources

Pingd connects directly to your existing databases and data platforms to power AI agent queries and advisor workflows. No data migration or syncing required — Pingd queries your data live, where it already lives.

Supported Data Sources

Data Source	Best For
Snowflake	Data warehouses with large-scale analytics data
PostgreSQL	Application databases and operational data
Redshift	AWS data warehouses
Amazon S3	CSV and JSON files stored in S3 buckets
Salesforce	CRM data (accounts, contacts, opportunities)

Before You Start

• You'll need Admin or Owner permissions in Pingd
• Have your database credentials ready (host, username, password, database name)
• Ensure your database allows connections from external IPs (you may need to whitelist Pingd's IP)
• For Salesforce, you'll need OAuth authorization

Using the Connection Wizard

Navigate to Dashboard → Connections and click New Connection to open the Connection Wizard.

Step 1: Choose Your Data Source Type

Select your database type from the list. Each type has its own credential fields.

Step 2: Enter Connection Details

Fill in the required credentials for your data source (see the sections below for what each type needs).

Step 3: Test Connection

Click Test Connection to verify Pingd can reach your database. The test checks:

• Network connectivity
• Authentication credentials
• Permission to list tables

If the test fails, double-check your credentials and network/firewall settings.

Step 4: Discover Tables

Once connected, Pingd automatically discovers the tables and schemas in your database. You'll see a list of available tables with their column definitions.

Step 5: Set as Primary

Click the menu (⋮) on your connection and select Set as Primary to make it the default data source for your AI agent and advisors. Each organization can have one primary connection.

---

Snowflake

Snowflake is ideal for organizations that centralize their data in a cloud data warehouse.

What You'll Need

• Account identifier — Your Snowflake account URL (e.g., skxzkvo-jw60957)
• Username and Password — A Snowflake user with read access
• Database and Schema — The database and schema containing your tables (e.g., PINGD_CRM.SALES)
• Warehouse — The compute warehouse to use for queries
• Role (optional) — A Snowflake role with the necessary permissions

Permissions Required

The Snowflake user needs at minimum:

• USAGE on the warehouse
• USAGE on the database and schema
• SELECT on the tables you want to query

Tips

• Use a dedicated read-only user for Pingd — don't share credentials with other services
• Choose a warehouse size appropriate for your data volume (X-Small works for most use cases)
• Column names in Snowflake are automatically handled regardless of case

---

Amazon S3

Connect CSV or JSON files stored in Amazon S3 buckets. Great for data exports, flat-file integrations, or supplementary datasets.

What You'll Need

• AWS Access Key ID and Secret Access Key — An IAM user with S3 read access
• Bucket name — The S3 bucket containing your files
• Region — The AWS region where your bucket lives (e.g., us-east-2)
• Path prefix (optional) — A folder path within the bucket (e.g., crm/)

Supported File Formats

• CSV — Comma-separated values with a header row
• JSON — Line-delimited JSON (one JSON object per line)

Permissions Required

The IAM user needs:

• s3:GetObject on the bucket/prefix
• s3:ListBucket on the bucket

Tips

• Keep files under 1 GB for best query performance
• Ensure CSV files have a consistent header row across all files
• Use folder prefixes to organize different data types

---

Salesforce

Connect your Salesforce CRM to query accounts, contacts, opportunities, and custom objects directly.

What You'll Need

Salesforce uses OAuth for authentication — no passwords to enter manually.

How to Connect

1. Click Connect Salesforce in the Connection Wizard
2. You'll be redirected to Salesforce to authorize Pingd
3. Log in with a Salesforce user that has API access
4. Grant the requested permissions
5. You'll be redirected back to Pingd with the connection established

Permissions Required

The Salesforce user needs:

• API Enabled permission
• Read access to the objects you want to query (Accounts, Contacts, Opportunities, etc.)

Supported Objects

Pingd can query any standard or custom Salesforce object your user has access to, including:

• Accounts
• Contacts
• Opportunities
• Leads
• Tasks
• Custom objects

Tips

• Use a dedicated Salesforce integration user rather than a personal account
• If your token expires, reconnect through the Connection Wizard — Pingd will re-authorize automatically

---

Redshift

Amazon Redshift connects just like PostgreSQL. Ideal for organizations using AWS data warehouses.

What You'll Need

• Host — Your Redshift cluster endpoint (e.g., my-cluster.abc123.us-east-1.redshift.amazonaws.com)
• Port — Usually 5439
• Database — The database name
• Username and Password — A Redshift user with read access
• Schema — The schema containing your tables (e.g., public)

Permissions Required

• USAGE on the schema
• SELECT on the tables you want to query

Tips

• Use a read-only user with access limited to the schemas Pingd needs
• Redshift uses the same PostgreSQL wire protocol, so if you're familiar with PostgreSQL setup, this will feel identical

---

PostgreSQL

Connect to any PostgreSQL database, including hosted services like Amazon RDS, Google Cloud SQL, Supabase, and Neon.

What You'll Need

• Host — Your database hostname or IP
• Port — Usually 5432
• Database — The database name
• Username and Password — A PostgreSQL user with read access
• Schema — The schema containing your tables (e.g., public)
• SSL — Whether to use SSL for the connection (recommended)

Permissions Required

• USAGE on the schema
• SELECT on the tables you want to query

Tips

• Use a read-only user — Pingd only needs SELECT permissions
• If your database is behind a firewall, add Pingd's IP to your allowlist
• For connection poolers (like PgBouncer or Supabase pooler), use the pooler's connection string

---

After Connecting

Once your data source is connected and set as primary:

1. Map table names — Pingd discovers your tables automatically. You can set logical names (like "accounts") to map to your physical table names for easier querying.
2. Set up column mappings — If your column names differ from standard names, configure mappings so Pingd understands your schema.
3. Configure data access policies — Set who can see what data through the Data Access Policies page.
4. Create advisors — Start building Advisors that analyze your connected data.
5. Ask questions — Your AI agent can now answer questions about your data via Slack.

Managing Connections

From Dashboard → Connections, you can:

• View all connections with their status (connected, error, pending)
• Test a connection to verify it's still working
• Set a primary connection for agent queries and advisors
• Edit credentials if passwords or tokens change
• Delete connections you no longer need

Troubleshooting

Connection test fails

• Verify your credentials are correct
• Check that your database allows external connections (firewall rules, IP allowlists)
• Ensure the database user has the required permissions
• For Snowflake, confirm the warehouse is running and not suspended

Tables not appearing

• The connected user may not have SELECT permission on those tables
• Try running table discovery again from the connection menu
• Check that you're looking at the correct schema

Queries timing out

• Your database may need performance tuning for the queries being run
• Consider adding indexes on frequently queried columns
• For Snowflake, try using a larger warehouse size
• Check if your data source has query timeout limits configured

For more help, see the Troubleshooting Guide.