Zoho rolls out in-house Zia LLM, strengthens enterprise AI capabilities

SaaS major Zoho launched its in-house large language model, Zia LLM, as part of its broader AI push at Zoholics India, its annual user conference held in Bengaluru, on Thursday.

Built entirely on NVIDIA’s AI-accelerated computing platform, the model addresses business-specific use cases, such as data extraction, summarisation, retrieval-augmented generation (RAG), and code generation.

The newly launched LLM comprises three models with 1.3 billion, 2.6 billion, and 7 billion parameters, each trained separately and tuned for different contexts, Zoho said in a statement, adding the models perform competitively against similar open-source LLMs.

"Today's announcement emphasises Zoho's longstanding aim to build foundational technology focused on the protection of customer data, breadth and depth of capabilities because of the business context, and value," said Mani Vembu, CEO of Zoho. 

He added, "Our LLM model is trained specifically for business use cases, keeping privacy and governance at its core, which has resulted in lowering the inference cost, passing on that value to the customers, while also ensuring that they are able to utilise AI productively and efficiently." 

Zoho said the three-model architecture helps balance performance and compute efficiency based on use case needs. It plans to expand Zia LLM’s capacity, with larger models expected to roll out by the end of 2025.

At present, Zia LLM is deployed across Zoho’s data centres in the US, India, and Europe. While the company continues to support integrations with other LLMs, like ChatGPT and Llama, its model allows users to keep their data on the firm’s own servers, avoiding external cloud providers. 

The model is presently tested internally across Zoho’s application suite and will be made available to customers in the coming months.

The company also introduced English and Hindi Automatic Speech Recognition (ASR) models, a no-code agent builder called Zia Agent Studio, and a Model Context Protocol (MCP) server, which enables third-party agents to access Zoho’s application actions.

Zoho said its two proprietary ASR models for speech-to-text conversion are built to run on low computing power while maintaining high accuracy. The models also claim to perform up to 75% better than similar models in standard tests. It plans to add support for more Indian and European languages and will also launch a reasoning language model (RLM) soon.

The Chennai SaaS giant has also launched over 25 ready-to-deploy Zia Agents, including a few specifically for India—one of its top markets—where Zoho's business grew by 32% in 2024. 

In India, top products by revenue were Zoho One (operating system for business), Workplace (enterprise email and collaboration suite), CRM Plus (customer experience platform), CRM, Books (accounting software) and Zoho People (HR management). ITeS, BFSI, manufacturing, retail, and education sectors drove this growth. 

It reported a 10-year compound annual growth rate (CAGR) of 51% in India.