Zoho has updated its business intelligence platform, Zoho Analytics, with semantic and ontology layers, a multi-agent architecture for Ask Zia, additional data connectors and expanded governance controls to support AI-driven analytics and workflows.
The semantic layer defines business terms and metrics, while the ontology layer maps organisational structures and relationships. Together, they provide AI agents with business context to interpret data and respond to queries.
Ask Zia, Zoho Analytics’ conversational business intelligence assistant, now uses six specialised sub-agents for data analysis, modelling, visualisation, data preparation, coding and SQL generation. Businesses can also build custom agents and connect them to internal and external systems through the Model Context Protocol (MCP). A workflow builder allows users to design automated processes based on data and analytics.
Zoho said its native semantic and ontology layers allow businesses to create and edit multiple semantic models without code or a separate cloud data warehouse.
The platform is adding 30 data connectors to its existing portfolio of more than 500. These include Zoho ERP, Zoho Spend, Zoho FSM, Zoho Shifts and Zoho Voice, alongside third-party services such as SAP, PayPal, Adobe Commerce, Twilio, Notion, Intercom, Supabase, Amazon DocumentDB and AlloyDB.
For developers, Zoho Analytics supports a headless BI model through its MCP server, available in remote and local configurations, with tools for data analysis, preparation and visualisation. The update also includes a command-line interface and a plug-in for accessing Zoho Analytics from within Claude.
Businesses can embed agents into their own applications, portals and products through white-labelled implementations, with permissions and governance controls retained outside the Zoho platform.
Governance features include a Data Catalog for managing data, metrics, analytics assets and models, with tools for data lineage, impact analysis and access policies. AI observability allows administrators to monitor agent activity, users, data access and responses. The platform also provides AI usage and cost tracking across teams.
Organisations can choose between Zoho’s large language model and other frontier models for Ask Zia. Sandbox and deployment tools allow teams to test analytics assets and agents before publishing them to production and to roll back to earlier versions when required.
Other updates include new visualisations, custom visuals, pixel-perfect reports, scenario analysis, a correlation matrix and a VARX forecasting model. The platform also adds a data science studio with Jupyter notebooks and PMML import, along with an updated mobile interface for Ask Zia.