Analyst(s): Keith Kirkpatrick
Publication Date: July 17, 2025
What is Covered in this Article:
- Proprietary, Business‑Tuned Zia LLM & ASR Models: Zoho introduced its own Zia LLM in three sizes, each optimized for specific tasks, and two in‑house ASR models for English and Hindi speech‑to‑text.
- Embedded & Prebuilt AI Agents with No‑Code Studio: The platform now includes a lineup of native Zia Agents, while Zia Agent Studio now provides a fully prompt‑based, no‑code (with optional low‑code) environment and over 700 built‑in actions for rapid agent creation and deployment.
- Agent Marketplace & Ecosystem Expansion: Zoho launched an Agent Marketplace within its broader Zoho Marketplace, featuring 25+ ready‑to‑deploy agents and supporting partners and developers in publishing their own custom agents.
- Interoperability & Future Roadmap for Agentic AI: Through its MCP server, Zoho exposes action libraries from 15+ applications for third‑party agents under strict permission controls; upcoming plans include scaling Zia LLM to larger sizes, expanding speech‑to‑text languages, introducing a Reasoning Language Model, and enabling agent‑to‑agent collaboration.
The News: Zoho Corporation announced several AI capabilities in mid-July, which revolve around enhancing the functionality of AI and AI agents within and outside of the Zoho platform. The key announcements center on the company’s new, proprietary Zia LLM, the company’s context-aware Zia Agents (with 25+ prebuilt agents), the no‑code Zia Agent Studio, and a Model Context Protocol (MCP) server that exposes Zoho’s action library to third‑party agents.
Zia LLM: Business‑Focused Large Language Models: Zoho announced a new, in-house large language model – Zia LLM – which is built on NVIDIA’s AI‑accelerated platform and comprises three parameter‑scaled variants—1.3 B, 2.6 B, and 7 B—each of which are tuned for specific business tasks such as structured data extraction, summarization, retrieval‑augmented generation (RAG), and code generation.
ASR Models: Zoho has developed two proprietary automatic speech recognition (ASR) models for English and Hindi speech‑to‑text conversion, with support for additional languages is slated for future releases.
New Native AI Agents: Zoho has embedded a roster of native AI agents directly into its applications, including a new version of Ask Zia, a platform‑wide conversational assistant now enhanced with BI‑focused skills and Customer Service Agent for Zoho Desk, which is capable of triaging or resolving support inquiries by understanding incoming requests, handling routine issues autonomously, and escalating complex cases to human agents to accelerate resolution times.
Zoho also announced updates to Zia Agent Studio, which now offers a fully prompt‑based, no‑code experience (with optional low‑code hooks) and grants access to over 700 built‑in actions spanning the Zoho product suite. Users can deploy these agents autonomously—via a button click, rule‑based triggers, or within live customer engagements.
Zoho also announced the launch of Agent Marketplace, housed within the broader Zoho Marketplace ecosystem. The platform includes more than 25 ready‑to‑deploy agents and will soon allow ecosystem partners, ISVs, and individual developers to publish their own. Some early agents include a revenue growth specialist, a deal analyzer, and a candidate screener.
Announced Interoperability via Model Context Protocol (MCP): Zoho said its MCP server exposes action libraries from over 15 Zoho applications (with Zoho Flow integrations available) to any MCP‑compliant client, enabling third‑party agents to access data and perform actions in a way that fully respects user‑defined permission schemes.
Roadmap & Future Enhancements: Zoho said that in the near future, it plans to scale Zia LLM with larger parameter configurations before the end of 2025 and will broaden speech‑to‑text language coverage across Europe and India. Later releases will introduce a Reasoning Language Model (RLM), add specialized skills to Ask Zia for finance and customer‑support teams, and implement Agent‑to‑Agent (A2A) interoperability so that Zia Agents can collaborate both with one another and with agents on other platforms.
Zoho Unveils Zia LLM, No‑Code Agent Studio, and Open Agent Interoperability
Analyst Take: Zoho has long been focused on delivering enterprise-scale products and services at price points that were easy for mid-market and small businesses to digest. The company’s recent announcements around enhanced AI functionality continue to support this approach and are designed to ensure that the vendor continues to provide the performance and capabilities desired by the market.
In particular, announcing a proprietary LLM is interesting, enabling Zoho to accomplish a few client-centric goals. For one, it can ensure that all data passing through the LLM is properly rail-guarded and constrained, a key concern for all organizations using generative AI. Zoho says the models’ performance figures are generally on par with other commercially available and open-source models, and they expect performance to improve as they refine and improve the models.
In addition, the release of several models specifically scaled and tuned for tasks should ensure greater precision and efficiency, which will help Zoho constrain the cost for customers using the model. While a limited number of tokens will be available within the base license price, Zoho expects to utilize a traditional consumption model for customers as they scale the technology across the business.
Moving Forward with Purpose-Built Agents
AI agents are quickly becoming table stakes for enterprise-focused SaaS vendors, and Zoho’s announcements of new pre-built and configured Zia agents, an agent marketplace, and updates to Zia Agent Studio are checking all the boxes for ensuring that organizations can create, launch, and manage agents easily and securely. Like other vendors, Zoho would like to position its platform as the control center for its own agents, as well as agents developed by an organization’s in-house development staff, as well as other vendors’ agents, and its support of Model Context Protocol (MCP) and Agent2Agent (A2A) interoperability protocols are key levers in helping to achieve that goal.
The most interesting part of the news is the future roadmap plan to introduce a reasoning language model. This model will further help agents understand context and semantic meaning, improving accuracy and efficiency when using generative AI and agents. This will be critical as customers increasingly focus on agentic accuracy and cost as they evaluate agentic platforms.
All told, Zoho’s announcements reflect a vendor tuned into its customers’ needs and desires. These customers are prioritizing AI and agentic technology to improve customer and employee experiences, drive productivity, and uplevel efficiency. The challenge for Zoho will be keeping pace with other competitors, particularly model-development companies, which continue to innovate and do not need to balance investment priorities across a wide range of products.
The right-sizing and purpose-built approach to offering different parameter models is a good step in ensuring models are deployed efficiently. To engender customer trust, Zoho will need to demonstrate its commitment to model safety and governance.
What to Watch:
- Zoho’s move to build and govern its own Zia LLM (in three task‑tuned sizes) guarantees data control, regulatory compliance, and performance parity with third‑party offerings. This may push other vendors to invest in their own LLM stack—or strengthen partnerships—that deliver similar guardrails without ballooning R&D and infrastructure costs.
- By offering 1.3 B, 2.6 B, and 7 B‑parameter variants, Zoho enables customers to match model complexity to task requirements and budget constraints. Competitors should watch how this “right‑sizing” approach impacts customer expectations around pricing transparency, token‑based consumption, and overall TCO for generative AI services.
- With Zia Agent Studio’s prompt‑only, no‑code UX (plus 700+ built‑in actions) and 25+ ready‑to‑deploy agents in its marketplace, Zoho is lowering the bar for non‑technical users to spin up intelligent agents. Rival platforms will need to bolster their low‑code/no‑code tooling and preconfigured agent libraries or risk losing mid‑market and SMB accounts.
- The introduction of a Model Context Protocol (MCP) server and forthcoming Agent‑to‑Agent (A2A) interoperability puts Zoho at the center of a growing multi‑agent ecosystem—while maintaining strict permission controls. Competing vendors must decide whether to adopt open protocols or double down on proprietary integrations to retain ecosystem partners and prevent customer lock‑in.
You can read the full press release providing additional information on all of these initiatives on the company’s website.
Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.
Other insights from Futurum:
Is Zoho’s Rural R&D Bet in Kerala a Scalable Model for Deep-Tech Innovation?
Can Zoho’s Unified Payment Stack Streamline the Order-to-Cash Journey?
Is Zoho’s Zia the Missing Link for Company-Wide CRM Adoption?
Author Information
Keith Kirkpatrick is Research Director, Enterprise Software & Digital Workflows for The Futurum Group. Keith has over 25 years of experience in research, marketing, and consulting-based fields.
He has authored in-depth reports and market forecast studies covering artificial intelligence, biometrics, data analytics, robotics, high performance computing, and quantum computing, with a specific focus on the use of these technologies within large enterprise organizations and SMBs. He has also established strong working relationships with the international technology vendor community and is a frequent speaker at industry conferences and events.
In his career as a financial and technology journalist he has written for national and trade publications, including BusinessWeek, CNBC.com, Investment Dealers’ Digest, The Red Herring, The Communications of the ACM, and Mobile Computing & Communications, among others.
He is a member of the Association of Independent Information Professionals (AIIP).
Keith holds dual Bachelor of Arts degrees in Magazine Journalism and Sociology from Syracuse University.
