Zoho Launches Catalyst 3.0 With Native Support for AI Coding Agents

Zoho has released Catalyst 3.0, a new version of its cloud platform for building and deploying applications, where the main audience for the changes is no longer just people writing code, but AI agents doing it instead. The company has added mechanisms designed to move the process straight from a prompt in a tool like Claude Code or Codex to a production environment, without manually stitching together a chain of cloud services.

What Exactly Changes in Catalyst

Catalyst has for years functioned as Zoho's PaaS (Platform-as-a-Service), competing with solutions like Firebase or AWS Amplify, offering databases, serverless functions, hosting, and integrations in one place. Version 3.0 adds a layer designed specifically for AI agents, rather than for a developer clicking through a dashboard.

Agent Skills is a set of instructions that let coding assistants understand Catalyst's architecture, SDK patterns, and platform best practices well enough to write code that's ready for deployment right away. Instead of generating code a developer then has to adjust to fit Catalyst's specific APIs, the agent is meant to already know the platform's constraints and conventions.

An Unattended CLI and an MCP Server

The second piece is a non-interactive command-line interface that can carry out multi-step deployment operations without asking for confirmation at every step. That only makes sense when an agent is driving the process rather than a human, since for a developer working manually, the absence of checkpoints tends to be a risk rather than a convenience.

The Model Context Protocol server, meanwhile, gives direct access to Catalyst services from within environments like VS Code or Claude Code, without switching to a separate console. In recent months MCP has become the de facto standard for connecting AI agents to external tools and systems, and Zoho joins the list of companies rolling out their own MCP server instead of waiting for third-party integrators to build one.

Governance as an Answer to Autonomous Deployment Risk

Giving AI agents the ability to deploy code to production on their own raises an obvious question about control. Zoho answers it with a set of governance mechanisms: separating development environments from production, limited permissions for collaborators, and full audit trails covering application logs, platform logs, and MCP tool calls. That's a meaningful signal, because past coding-agent mishaps, from accidental data deletion to vulnerabilities introduced by automated deployments, have shown that an agent's speed without an audit trail can get expensive.

With an agentic, serverless full-stack platform, Catalyst gives developers and AI agents complete cloud independence - Anand Nergunam, Vice President of Revenue Growth, Zoho

With Catalyst, developers can leverage the platform's agentic skills, MCP server, and built-in orchestration mechanisms to deploy - Jason Bloomberg, Managing Director, Intellyx

Who This Change Is For

Zoho is targeting three audiences: systems integrators and software firms building solutions for clients, independent developers, and students, for whom the platform is entirely free, with no payment card required. Among the customers testing the new features, the company lists MAP Communications and Eternia by Hindalco, suggesting the tool is already reaching corporate use cases and not just hobby projects.

For Polish software companies and freelancers, what matters is that Catalyst 3.0 lowers the barrier to entry for building products based on coding agents, by offering a free starter tier with credits spread over six months. It's a model similar to what Vercel or Supabase used earlier, to attract developers experimenting with new AI tools before they commit to a paid plan.

A Broader Trend in Developer Tools

Zoho's move fits a broader trend in which cloud and developer platform providers are racing to integrate with coding agents rather than build their own language models from scratch. Other companies have taken a similar path in recent months, adding support for Claude Code, Codex, or Gemini CLI directly into their API gateways and data platforms. For Zoho, a company known mainly for its office suite and CRM, an agentic Catalyst is an attempt to stake a claim in the developer infrastructure segment before larger cloud players come to dominate the market.

It remains an open question whether the governance Zoho has proposed will hold up in practice once an AI agent has full access to deploy code to production without step-by-step oversight. Other companies' experience with autonomous agents so far shows that audit logs help determine what went wrong, but don't always prevent the incident itself.