← All updates Databricks Update One-Pager Light version
databricks Platform · Updated Jun 25, 2026

Genie Agent Enhancements

New authoring skills for Genie Agents improve usability and accuracy.

What it does

The Genie Agent now includes enhanced authoring skills that allow users to create and maintain agents more effectively. Users can test agents with real-world questions, review generated SQL and visualizations, and provide feedback to improve accuracy. Additionally, the system supports monitoring agent usage and user feedback to refine responses over time.

Why it matters

These enhancements enable data teams to create more accurate and responsive Genie Agents, leading to improved user satisfaction and productivity. By allowing for real-time feedback and adjustments, organizations can ensure their agents remain relevant and effective as business needs evolve, ultimately driving better decision-making.

Key capabilities
Testing with Real Questions
Users can interact with the Genie Agent by asking realistic questions to evaluate its performance.
Response Feedback Mechanism
Users can provide feedback on responses, allowing for continuous improvement of the agent's accuracy.
SQL Query Review and Editing
Authors can review and edit generated SQL queries to correct inaccuracies and save improvements.
Usage Monitoring
The Monitor tab allows users to track agent performance, user interactions, and feedback trends.
Prerequisites
Users must have appropriate permissions (e.g., CAN MANAGE, CAN EDIT) to access all features.
Genie Code must be enabled for debugging and context updates.
Recommended use cases
Data teams can refine agents based on user feedback to improve accuracy.Business users can interact with agents to obtain insights from data without needing technical expertise.Organizations can monitor agent performance over time to identify areas for improvement.
Limitations · remember
!Responses may vary due to the non-deterministic nature of large language models.
!Conversations created before enabling Genie Chat Sharing remain private.
Architecture
Genie Agent → Genie Code → Monitoring Interface → User Feedback Loop
Get started
1
Create a Genie Agent
Follow the setup instructions to create a new Genie Agent.
2
Test the Agent
Use realistic questions to evaluate the agent's responses.
3
Provide Feedback
Respond to agent outputs to flag inaccuracies and suggest corrections.
4
Review SQL Queries
Inspect and edit generated SQL queries as needed for accuracy.
5
Monitor Usage
Utilize the Monitor tab to track interactions and feedback trends.
Key takeaway
The enhancements to Genie Agents empower users to create more accurate and responsive data-driven solutions.
2026-06-25-genie-agent-formerly-genie-space-enhancements Generated by the Databricks Updates Agent · browse all updates