Skip to main content
No additional setup needed! If you’ve already connected data sources in Julius, they’re immediately available through the Slack agent.

​
How Automatic Detection Works

Julius uses intelligent context analysis to determine which data connector to use:
  1. Query Analysis: Julius reads your question and identifies data references
  2. Connector Matching: Matches keywords against your configured data connector names and types
  3. Automatic Connection: Connects to the appropriate data source without requiring explicit instructions
  4. Smart Fallback: If ambiguous, Julius asks you to clarify which connector to use

​
Best Practices for using data connectors in Slack

To make automatic detection work smoothly, it’s important to name your data connectors clearly, so you can reference them from Slack easily.

​
Connector Naming Conventions

How you name your connectors directly impacts how easily Julius can detect and use them in Slack queries. Follow these guidelines when naming your data connectors:

​
✅ Do’s

Use the environment or purpose in the name:
  • prod-read-replica - Clearly indicates this is a production read-only copy
  • analytics-warehouse - Shows the primary use case
  • staging-db - Indicates it’s a staging environment
  • customer-data - Describes the data type it contains
Keep names short and memorable:
  • prod-replica - Short and easy to reference
  • analytics - Simple one-word name works well
  • main-db - Descriptive but concise
Be specific to avoid ambiguity:
  • prod-read-replica-postgres - Very clear what this is
  • west-coast-warehouse - Specific enough that there’s no confusion with other connectors

​
❌ Don’ts

Don’t name connectors after the database type alone:
  • ❌ postgres - Too generic, especially if you have multiple Postgres databases
  • ❌ bigquery - Doesn’t distinguish between different BigQuery projects
  • ❌ snowflake - Vague if you have multiple Snowflake connections
Don’t use ambiguous or vague names:
  • ❌ db - Too vague, Julius won’t know which connector you mean
  • ❌ main - Ambiguous across multiple teams or projects
  • ❌ database - Not descriptive enough
  • ❌ data - Doesn’t help Julius identify the right connector
Don’t use names that are too similar to other connectors:
  • ❌ postgres-1, postgres-2 - Hard to remember which is which
  • ❌ db-prod, db-production - Confusingly similar
  • ❌ analytics-v1, analytics-v2 - Ambiguous which version to use

​
Example Usage:

Imagine you have a Postgres connection named “prod read replica” in your Julius account. Let’s walk through good and bad examples of referencing this connector in your queries:

​
✅ Good Examples

Use the exact connector name: 
@Julius using prod read replica, get the most recent users from the past week
Julius will immediately connect to your Postgres database and find recent users. Mentioning the database type : 
@Julius using Postgres, get the most recent users from the past week
Julius will be able to detect you’re talking about the “prod read replica” connector.
If you have multiple connectors of the same type, Julius will have a hard time disambiguating between them, and may get it wrong. Use the connector name instead.
Include the purpose: 
@Julius using our read replica, show me the top 10 customers by spend
Julius recognizes “read replica” from your connector name and connects automatically.

​
❌ Bad Examples

Vague query without connector reference: 
@Julius get the most recent users from the past week
Julius won’t get a good sense for which connector to use. If you only have one connector, this query may work ok. Wrong connector type: 
@Julius using BigQuery, get recent users
You only have a Postgres connector, not BigQuery. Ambiguous reference without enough context: 
@Julius using db, show me everything
“db” is too vague. Julius can’t match this to “prod read replica” reliably. Be more specific:

​
Query Examples

Here are common query patterns that work well with automatic connector detection:
  • Explicit Connector
  • Database Type
  • Natural Language
  • Multiple Joins
@Julius using prod-replica, what were our daily signups for October?
Julius will use the connector named “prod-replica”

​
Troubleshooting Connector Issues

Cause: No data connectors are configured in your Julius account.Solution:
  1. Go to julius.ai/settings/data_connectors
  2. Click “Add New Connector”
  3. Follow the setup guide for your database type
  4. Test the connection before using in Slack
Cause: The connector name might be spelled differently than expected.Solution:
  • Check the exact name of your connector in Julius settings
  • Use the database type instead (e.g., “Postgres” instead of specific name)
  • Ask Julius to list available connectors: “@Julius what data sources can you access?”
Cause: The database connection is misconfigured or credentials are invalid.Solution:
  1. Test the connector directly in Julius web app
  2. Verify the database credentials are correct
  3. Check that the database user has read permissions
  4. Ensure IP whitelisting is configured if your database requires it
Cause: The connector name is ambiguous or matches multiple connectors.Solution:
  • Specify the full connector name: “@Julius using [exact-name], …”
  • Rename the connector to be more unique
  • Use the database type for clarity

​
Next Steps

Setup Data Connectors

Configure your first data connector in Julius

Query Best Practices

Learn best practices for writing effective queries in Slack

Scheduled Reports

Automate recurring analysis and data reports to Slack
Questions about your connectors? Check our Data Connectors Guide or reach out to team@julius.ai.