Skip to main content

Data Agents Overview

Autonify's Data Agents provide deep research capabilities powered by AI to automatically analyze your data and build comprehensive business intelligence reports and executive packs. These agents can identify revenue opportunities, data quality issues, compliance risks, and operational inefficiencies across your entire data estate.

VIDEO TUTORIAL⏱️ 4 mins

📹 Data Agents Deep Dive

Learn how to leverage AI-powered data agents to build business intelligence reports and identify opportunities across your data.

Overview​

Data Agents are AI-powered analytical tools that perform deep investigation of your connected databases. They execute multi-phase research workflows to understand your data structure, identify patterns, and generate actionable business insights with quantified monetary impact.

Agent Packs​

Agent Packs are pre-configured collections of workflows designed for specific business objectives. Each pack contains multiple workflows that work together to provide comprehensive analysis of your data.

Pre-built Agent Packs​

Autonify provides several pre-built agent packs tailored to different industries and use cases:

  • Revenue Optimization Pack - Identify revenue opportunities, pricing inefficiencies, and growth potential
  • Real Estate Intelligence - Specialized analysis for property management and real estate operations
  • Banking Intelligence - Financial services focused analysis including compliance and risk assessment
  • Data Sensitivity - Identify and classify sensitive data across your organization

Pack Workflows​

Each agent pack contains multiple specialized workflows. For example, the Revenue Optimization Pack includes 12 workflows:

  • Top Revenue Opportunities - Identify high-value revenue opportunities in your data
  • Churn Prevention - Detect patterns indicating customer churn risk
  • Cross-sell Opportunities - Find potential cross-selling opportunities
  • Pricing Opportunities - Analyze pricing inefficiencies and optimization potential
  • Sales Efficiencies - Identify areas to improve sales operations
  • Revenue Leakage - Detect revenue loss from operational issues
  • And more specialized workflows

Creating Custom Agent Packs​

You can create custom agent packs tailored to your specific business needs.

Steps to Create an Agent Pack​

  1. Navigate to the Data Agent Library in your data source
  2. Click the Create Agent button in the top right
  3. In the dialog, provide:
    • Name - A descriptive name for your agent pack
    • Description - What this agent pack will analyze
    • Industry (optional) - Select from: General, Healthcare, Finance, Retail, Manufacturing, Technology, Education, Government, Real Estate, Transportation, Energy, or Media & Entertainment

Create Agent Dialog Create a custom agent pack with name, description, and industry selection for tailored analysis

Adding Workflows to Agent Packs​

After creating an agent pack, you need to add individual workflows (called audits) that perform specific analyses.

Creating Workflows​

  1. Open your agent pack from the Data Agent Library
  2. Click Add Workflow button
  3. Configure the workflow with:
    • Name - Specific analysis name (e.g., "Top Revenue Opportunities")
    • Description - Detailed explanation of what this workflow analyzes
    • Priority - Critical, High, Medium, or Low
    • Prompt - The specific instructions for the AI to execute
    • Required Tools - Select which tools the workflow can access

Available Tools​

Workflows can be configured with access to different analysis tools:

  • Data Catalog - Access to complete data catalog and metadata
  • Data Integration - Direct query access to analyze your data
  • Compliance Engine - PII detection and sensitivity classification
  • Data Quality - Access to quality metrics and validation rules

Add Workflow Dialog Configure workflow settings including name, priority, prompt, and tool access permissions

Running Agent Workflows​

Executing a Workflow​

  1. Open an agent pack and navigate to the workflow
  2. Click the three-dot menu on the workflow card
  3. Select Run Workflow from the dropdown
  4. The agent begins its multi-phase investigation

Workflow Card Dropdown Access workflow actions through the dropdown menu to run, edit, or manage workflows

Monitoring Progress​

While the agent runs, you can monitor its progress through:

  • Real-time status updates in the progress dialog
  • Phase completion indicators showing current analysis stage
  • Percentage complete and progress messages
  • Option to minimize to continue working while agent runs

Understanding Agent Results​

Investigation Findings​

After completion, the agent presents its findings organized by:

Severity Levels​

Findings are categorized by severity:

  • Critical - Immediate action required, significant business impact
  • High - Important issues requiring prompt attention
  • Medium - Notable findings for consideration
  • Low - Minor observations and optimizations

Monetary Impact​

Each finding includes:

  • Potential Value - Estimated monetary impact (e.g., £900,000)
  • Risk Assessment - Financial exposure from identified issues
  • Opportunity Sizing - Revenue potential from recommendations

Findings List View analysis findings organized by severity with estimated monetary impact for each issue

Detailed Finding Analysis​

Each finding is expandable and contains:

Finding Components​

  • Title - Clear description of the issue or opportunity
  • Description - Detailed explanation of the finding
  • Severity Badge - Visual indicator of impact level
  • Financial Impact - Estimated monetary value (if applicable)

Evidence Section​

  • Specific data examples and patterns identified
  • Query results supporting the finding
  • Statistical analysis backing the conclusion

Recommendations Section​

  • Actionable steps to address the finding
  • Implementation guidance
  • Expected outcomes from remediation

Expanded Finding Detailed finding view showing evidence, specific data examples, and actionable recommendations

Example Finding: Duplicate Transactions​

Issue: High-volume duplicate transactions overstating revenue

Evidence:

  • Identified 15,000 duplicate invoice records
  • Same transaction IDs appearing multiple times
  • Pattern detected across Q3 and Q4 data

Recommendations:

  1. Implement unique constraint on transaction_id column
  2. Create data quality rule to prevent duplicates
  3. Run deduplication process on historical data
  4. Update ETL pipeline to include duplicate checks

Potential Impact: £450,000 in overstated revenue

Executive Reports​

Report Generation​

Workflow results can be downloaded as PDF reports:

  1. After a workflow completes, navigate to the results
  2. Click the Download Report button
  3. The PDF includes:
    • Executive Summary - High-level overview of findings
    • Detailed Findings - All issues with evidence
    • Financial Impact - Total monetary impact assessment
    • Recommendations - Prioritized action items
    • Technical Details - Supporting data and queries

Download Report PDF Export comprehensive executive reports as PDF documents for sharing and presentation

Report Sharing​

Generated reports can be:

  • Downloaded as PDF documents
  • Shared with executive teams
  • Used for board presentations
  • Integrated into compliance documentation

Best Practices​

Workflow Selection​

  • Start with high-priority workflows aligned to business objectives
  • Run complementary workflows together for comprehensive analysis
  • Schedule regular runs to track improvements over time

Configuration Optimization​

  • Begin with default settings for initial runs
  • Adjust based on data volume and complexity
  • Use accuracy focus for critical business decisions
  • Apply speed focus for routine monitoring

Result Management​

  • Review critical findings immediately
  • Create action items from recommendations
  • Track remediation progress
  • Re-run agents to verify improvements

Use Cases​

Revenue Optimization​

  • Identify pricing inefficiencies across product lines
  • Detect revenue leakage from operational issues
  • Find cross-sell and upsell opportunities
  • Prevent customer churn through pattern analysis

Data Quality Improvement​

  • Discover data inconsistencies and duplicates
  • Identify missing or incomplete data
  • Validate business rules and constraints
  • Monitor data quality trends over time

Compliance and Risk​

  • Detect sensitive data exposure
  • Identify compliance violations
  • Assess data governance risks
  • Generate audit documentation

Operational Excellence​

  • Find process inefficiencies
  • Identify cost reduction opportunities
  • Optimize resource allocation
  • Improve decision-making with data insights

Technical Implementation​

Three-Phase Investigation Process​

The BI agent uses a sophisticated three-phase approach:

  1. Discovery Phase - Initial exploration of data structure and relationships
  2. Assessment Phase - Deep analysis of identified patterns and issues
  3. Evidence Gathering - Collection of specific examples and quantification

Iteration Limits​

To ensure efficient execution, agents have built-in safeguards:

  • Maximum 120 iterations per workflow execution
  • Automatic optimization based on data volume
  • Smart sampling for large datasets

Next Steps​