AI Assistant

This KB article is part of SYNCHUB's data-extraction knowledge base for Katana

Getting started
First, let's get your data
Your connection hub
Start reporting
Reporting in...
Power BI
Tableau
Excel
Zoho Analytics
Other BI Tools
Microsoft Access
Qlik
ODBC Connection
Query Editor
Looker (formerly Google Data Studio)
Insights
Supabase
Going deeper
Connecting to Xero
Connecting to WorkflowMax
How it works
Your connection dashboard
Importing your historical data
Data security
Data structure
Syncing specific data segments
Viewing logs
Trails
Connecting to Cin7
Bring your own database
Connecting to HubSpot
Keeping your data secure
Restricting IP access via your database firewall
Datastore management
Data Model Explorer
Connecting to our MCP server
Connecting to Zoho CRM
Third-party integration
Integrating with n8n
Integrating with OpenAI AgentKit
AI Assistant
Data Accessibility
Data accessibility
Date modification filters
Paging
Nested data structures
Single vs Summary Records
Account
Billing
Users
Slack integration
Error codes
Invalid Remote ID
Invalid SQL
Our AI Assistant is a powerful tool lets you speak to your Katana data directly, with no BI or SQL skills required. ### How does it work? As part of extracting your data from the Katana API, the SyncHub team create a vast number of "meta" annotations about your data. For example, things like: - whether your dates are in UTC or local time - whether your items are hard deleted, or marked with an archived/deleted/etc flag This knowledge allows us to provide an LLM with an extremely accurate data model of your Katana database, and this in turn allows the LLM to map your questions to extremely accurate SQL queries. The basic process is: 1. you ask a question of your Katana data, using regular conversational language 2. the LLM converts this to a SQL query, which is passed back to our us 3. we take this SQL and execute it against your database (remember, your database is _read-only_, so there's no danger of damaging your data) 4. the pass the results of the SQL query back to the AI in the form of a CSV file 5. the AI uses this data to answer the original question ### Using the AI Assistant We provide various mechanisms for accessing the AI Assistant. #### Our AI Assistant First, we have a fully-operational chatbot of our own, which you can find under your main menu from within SyncHub. The AI Assistant comes with additional features beyond the data translation, including: 1. **Personalization & Memory.** Your assistant can remember nuances of your business which aren't necessarily stored in your data. Things like special customer needs, or business goals. 2. **Scheduled tasks.** Queue up recurring tasks and have our assistant run them automatically. For example, a daily sales summary, or an alert if employees don't show up to work on time. 3. **Charts.** If you're move of a visual person, our assistant will happily render charts to describe your data. 4. **Notifications.** Email or [Slack](/connectors/katana/knowledge-base/slack) notifications to other people in your organization. #### Connecting Slack to Katana Our [Slack integration](/connectors/katana/knowledge-base/slack) backs on the AI Assistant, allowing you to converse with your Katana data directly from within Slack. #### Use our MCP server for Katana We have built an [MCP](/connectors/katana/knowledge-base/mcp) wrapper around all of our data connections, allowing you to integrate your Katana data directly into your regular chatbots, like OpenAI or Claude.