Using Databases

Databases in Saltare are structured data stores that you can query with natural language. They work like smart spreadsheets — you define columns and rows, and AI agents can read, write, and analyze the data without you writing a single SQL query.

Creating a Database

1. Navigate to a database group in the sidebar
2. Click New Database
3. Give it a name (e.g., "Customer Feedback", "Vendor List", "Content Calendar")
4. Define your columns with names and types

Column Types

Databases support standard column types:
- Text — Free-form text content
- Number — Integers or decimals
- Boolean — True/false values
- Date — Calendar dates
- JSON — Structured data objects

Working with Data

The Table Interface

Databases render as a spreadsheet-like table. You can:
- Add rows by clicking the add button
- Edit cells by double-clicking
- Sort columns by clicking headers
- Scroll horizontally for wide tables

Row-Level Discussions

Each row has its own discussion thread. Click on a row to open the detail panel and start a conversation about that specific data point. This is useful for:
- Discussing a vendor's pricing with your team
- Flagging a data quality issue
- Asking an agent to research a specific entry

Natural Language Queries

The most powerful feature of Saltare databases is natural language querying. Instead of writing SQL, describe what you want in plain English:

@Assistant query the Customer Feedback database:
show me all entries from the last 30 days 
where sentiment is negative, sorted by date

The agent translates your request into a SQL query, executes it, and returns the results in a readable format.

Query Examples

Filtering:

@Analyst from the Sales Pipeline database,
show deals over $50,000 that are still open


Aggregation:

@Analyst what's the average deal size by region
in the Sales Pipeline database?


Comparison:

@Analyst compare this month's new leads vs last month
in the Lead Tracking database


Agent Data Operations

Agents can perform full CRUD operations on databases:

Creating Data

@Assistant add a new row to the Vendor List:
name "Acme Corp", category "SaaS", 
annual_cost 24000, status "evaluating"

Updating Data

@Assistant in the Vendor List, update Acme Corp's 
status to "approved" and add a note: "Contract signed 2024-03-15"

Bulk Operations

@Assistant import these companies into the Competitor Database:
- TechCo (Series B, 150 employees)
- DataFlow (Series A, 45 employees)  
- CloudBase (Seed, 12 employees)

Database Groups

Like tasks and documents, databases are organized into groups:
- Each group appears in the sidebar with a count of tables
- Groups have auto-created discussion channels
- Groups support archiving

Use Cases

CRM / Lead Tracking

Track prospects, deal stages, and contact information. Agents can update records from chat conversations.

Content Calendar

Plan and track content across channels. Agents can query upcoming deadlines and flag overdue items.

Research Databases

Store structured research data that agents populate during research tasks. Query results naturally.

Inventory / Asset Tracking

Track equipment, licenses, or any structured inventory. Natural language queries make reporting effortless.

Tips

• Define clear column names — Descriptive names help agents understand and query your data more accurately
• Use consistent data formats — Standardize how dates, currencies, and categories are entered
• Let agents populate data — Agents can bulk-create rows from research, saving manual data entry
• Combine with documents — Use database queries in documents for data-backed reports