Ask Zia - The Super Agent in BI
Generative AI is redefining how business users interact with data. In Zoho Analytics, every stage of the analytical workflow, from data preparation and visualization to building powerful machine learning models, is seamlessly infused with AI.
At the core of these capabilities is Ask Zia - Super Agent, a human-led, AI-assisted system that serves as the central orchestration layer of Zoho Analytics' Agentic AI framework. It coordinates seven specialized agents to understand analytical intent, execute tasks, and deliver intelligent recommendations and insights.
With a natural, conversational interface, it empowers users of different personas to build pipelines, complex queries, generate reports, insights, and perform actions, all through simple, intuitive prompts. This AI-driven assistant bridges the gap between data complexity and business decision-making with a human-in-the-loop.
Specialized AI Agents in Ask Zia
Ask Zia acts as an orchestration layer that brings together seven specialized AI agents to support different stages of the analytics workflow. Based on the user's analytical intent, the Super Agent identifies the appropriate capabilities, coordinates tasks across specialized agents, and delivers relevant results through a unified conversational interface.
- SQL & Formula Agent
- Dataprep Agent
- DSML Coding Agent
- Analysis Agent
- Visualization Agent
- Slideshow Agent
Access & Permissions
Users with Administrator privileges, Shared users and Custom roles users, can interact with Ask Zia and perform conversational analysis on their data, subject to the data access permissions defined for their role. Refer to the GenAI configuration article to learn more.
Ask Zia - Chat History
Ask Zia automatically saves your conversations, allowing you to access and revisit previous chats whenever needed. Select a conversation from the chat history to review earlier questions and responses or continue the analysis from where you left off. Chat History helps users maintain continuity in their data analysis, revisit earlier findings, and explore follow-up questions more efficiently. It reduces repetitive work and makes it easier to build on insights discovered during previous interactions.
Context Retention
Ask Zia maintains context throughout a conversation, understanding each query in relation to previous questions. This allows users to ask follow-up questions naturally without repeating the metrics or filters while continuing the conversation.
- Real-time type-ahead assistance suggests relevant fields such as columns, dimensions, measures, and functions, and may also recommend suitable visualizations or existing reports related to the query.
- Context-aware follow-up suggestions: During a conversation, Ask Zia analyzes the current query and proposes logical next steps, such as additional metrics, filters, or related analytical questions to refine the analysis.
- Proactive query recommendations: As users begin interacting with Ask Zia, it suggests relevant analytical questions by analyzing the data model, previous user interactions, and commonly used query patterns.
Best Practices for Using Ask Zia
- Start with simple queries: Begin with a high-level question and progressively refine it with follow-up queries to explore the data further.
- Use clear metric and dimension names: Use terms that closely match the column names in your dataset so Ask Zia can interpret the query accurately.
- Include relevant filters: Apply filters such as time periods, categories, or conditions to narrow the analysis to the most relevant data.
- Group metrics by dimensions: Analyze metrics across one or more dimensions to compare performance across categories or segments.
- Leverage follow-up queries: Use conversational context to refine the analysis without repeating the entire question.
- Specify time periods for trend analysis: Include time context such as months, quarters, or years when analyzing trends or performance over time.
- Gradually increase query detail: Add dimensions, filters, or sorting progressively to move from high-level summaries to deeper insights.
Embed Ask Zia
Embedding Ask Zia, the AI-powered analytics assistant, brings advanced analytical capabilities like conversational insights and Key Driver Analysis (KDA) directly into the applications users already work with. Users can ask business questions, explore data, detect patterns or anomalies, and identify the key factors driving their metrics without switching platforms. This makes data interaction more seamless, accessible, and insight-driven across the organization.
The GenAI Architecture
To power the AI capabilities, Zoho Analytics predominantly provides two LLMs - the Zoho LLM for report creation, and OpenAI LLM for capabilities such as data pipeline, report and dashboard creation, and insights generation.The administrators can choose which LLM to use based on business requirements and privacy.
Security, Privacy and Governance
At Zoho, we prioritize privacy from the ground up. For AI-powered features, the information shared with large language models (LLMs) is strictly limited based on the task type. In most cases, only prompts and metadata such as column names, data types, and the data model, are shared. Columns marked as Personally Identifiable Information (PII) are never shared under any circumstance.
For insight recommendations, only aggregated data is sent to the LLM to maintain both accuracy and confidentiality. You can manage and customize these data-sharing preferences in the GenAI Configuration settings page.