If you don’t know the new Catalyst, you don’t know Zoho
BrainBlog for Zoho by Jason Bloomberg
AI coding agents have unquestionably transformed software development. No longer do developers need to crank out code by hand. Instead, a simple prompt will suffice – and the coding agent will generate profuse volumes of code, supposedly ready to deploy.
For all but the simplest applications, however, building scalable, governed applications and deploying them into production is still too difficult.
The generated code is only one piece of the puzzle. Developers must assemble front-end and back-end components as well as supporting databases and security components and then choose one or more AI models – finally putting the whole thing into the proper cloud or other runtime environment.
Today, many vendors are tackling this AI coding deployment complexity from various angles, but one vendor stands alone in its approach: Zoho, with its revamped Catalyst platform.
Catalyst by Zoho has always been a Platform-as-a-Service (PaaS) offering, running in the same cloud infrastructure as the rest of Zoho’s apps.
Zoho has now rebuilt Catalyst as an AI agent-empowered platform that takes deployment complexity off developers’ plates, improving both productivity and the overall quality of production applications.
What’s inside the Catalyst agent-empowered PaaS
The PaaS terminology dates from the early days of the cloud. Many first-generation PaaS offerings struggled to deliver on their promise, largely because assembling all the infrastructure components necessary to run applications in scalable cloud infrastructure proved more difficult than expected.
Zoho’s long-term approach to building cloud infrastructure overcame these challenges. The company built its own bare metal data centers as well as the software runtime infrastructure to support its large portfolio of business applications.
As a result, all the various elements anybody could want from a PaaS, including databases, security infrastructure, real-time middleware, runtime environments, and all the horizontal scalability we’ve all come to expect from hyperscale clouds, are integral to the Zoho platform.
As a PaaS offering, therefore, Catalyst has always taken advantage of this tightly integrated, full stack context – giving developers the infrastructure they require to deploy applications at scale.
Fundamentally, Catalyst is a full-stack serverless cloud, where every component offers its own API and SDK, all running on Zoho’s global data center infrastructure.
Full stack PaaS – just add agents
Zoho has added three components to Catalyst to help developers turn AI-generated code into production-ready applications:
- The Catalyst Model Context Protocol (MCP) server for connecting AI coding agents to the full range of Catalyst’s capabilities. With this server, developers can execute common tasks in their IDEs like adding or changing database tables without switching to the Catalyst console.
- An automated ‘non-interactive’ command-line interface (CLI) enables AI coding agents to execute multi-step workflows in Catalyst without requiring human input at each step.
- Catalyst agent skills that guide coding agents toward whatever capabilities and workflows available in Catalyst are appropriate for the task at hand. By avoiding unnecessary capabilities, agent skills minimize token usage and enable developers to generate high-quality output, even with smaller AI models.
By combining these new capabilities with the rest of Catalyst, developers are able to take advantage of the platform’s built-in orchestration capabilities.
The Catalyst agent skills route requests to either the CLI or MCP path without expecting the AI model to make the decision, leading to more efficient tool choices and more reliable applications overall.
The underlying power of PaaS
Catalyst’s new capabilities are at the heart of the story, supporting developers as they leverage AI coding agents to build applications.
Those new capabilities are merely icing on the cake – while the cake itself is the Catalyst PaaS.
The Catalyst PaaS offers database and object storage capabilities, front-end modules, massively scalable compute resources, schedulers, and an event bus that ties everything together in real-time.
This infrastructure runs on the Zoho Platform – the same platform that supports Zoho CRM and dozens of other SaaS apps that millions of customers have been using for years. All running in Zoho’s own data centers on Zoho-configured hardware.
Catalyst also inherits Zoho’s comprehensive security and compliance infrastructure, including hardened authentication and authorization, DDoS protection, SOC compliance, regular vulnerability assessment and penetration testing (VAPT), a web application firewall, and full audit trails across application logs, platform logs, as well as MCP tool calling logs.
If you’re comparing Catalyst to other vendors’ various AI coding harnesses, be sure to keep this PaaS story in mind. Catalyst is far more than a harness. It’s the entire platform.
The Intellyx take
Today, developers can use Catalyst to build deterministic applications – applications that behave the same every time.
Support for building non-deterministic, agentic applications is on the Catalyst roadmap. Such applications are inherently unpredictable and thus require an entirely different approach to governance – an approach the industry at large is still trying to figure out.
Today’s Catalyst helps developers use AI coding agents – tools that themselves behave non-deterministically. Catalyst then gives such developers a straightforward path for deploying the resulting deterministic applications into production.
Catalyst threads this needle, helping developers use nondeterministic tooling to create deterministic applications – applications that drive business outcomes without going off the rails.
Copyright © Intellyx BV. Zoho is an Intellyx customer. Intellyx retains final editorial control of this article. A human wrote every word of this article. Image credit: Craiyon.
