Zoho Zia Agents are AI assistants that understand natural language, carry out tasks through API integrations, and draw on connected knowledge bases to provide accurate, context-aware responses. The Agent Studio is the central development environment where you design, configure, and launch your agents.

Before You Begin

To get the most out of Agent Studio, prepare the following resources before you start building:

• Behavioural guidelines outlining how the agent should respond and what topics it should address or avoid.
• API tools that enable the agent to perform the actions your workflows require.
• Reference documents such as product documentation, FAQs, policies, and internal guides the agent can draw on for accurate responses.
• Service connections providing authenticated access to the external systems the agent needs to interact with.

Two Ways to Create an Agent

Agent Studio offers two creation pathways: building from scratch for complete control, or using Zia AI for a faster, prompt-driven experience.

Option 1: Create From Scratch

This approach gives you precise control over every aspect of the agent. You work through four sequential configuration steps:

1. Basic Information. Give the agent a clear, function-based name and a purpose description. Choose your AI vendor (Zoho or OpenAI via ZKS, or a BYOK option such as Claude) and select a model appropriate to your use case complexity. Define the agent's role (for example, "Customer Insights Analyst") and write a comprehensive instruction set that covers the agent's expertise, expected behaviour, and any restrictions.
2. Knowledge Base Integration. Select the reference documents the agent should consult when formulating responses. You can attach multiple documents — load only what is directly relevant to keep responses focused and accurate.
3. Tool Integration. Add the tools the agent needs to take action. System tools are organised by Zoho service (CRM, Desk, Sheet, Calendar, and others); custom tool groups handle specialised functions. You can mix tools across service types and use the search function to locate specific tools quickly.
4. Guardrail Configuration. Built-in safeguards include fairness and bias monitoring and toxicity prevention. You can also define custom guardrails as specific Do's and Don'ts rules. Guardrail directives take precedence over general instructions, so they are the right place to set non-negotiable constraints.

Option 2: Create With Zia AI

The Zia AI pathway accelerates setup by generating a configuration from a natural language description of what you need. Quick-prompt templates are available for common roles including Sales Assistant, Support Agent, Technical Writer, and Product Manager. After you describe your requirements, Zia may ask follow-up clarification questions before producing a full configuration. The generated setup — including document and tool selections — is presented for your review and can be edited before you proceed.

Best Practices

• Write detailed instructions. Vague instructions lead to inconsistent behaviour. Be specific about what the agent should do, how it should respond, and what it should not do.
• Load only relevant knowledge. Attaching unnecessary documents dilutes the agent's focus. Curate your knowledge base carefully.
• Use guardrails for constraints. Any rule that must always be followed — regardless of how a user phrases their request — belongs in a guardrail, not just in the instruction set.
• Test before deploying. Use the built-in testing tools to validate the agent's behaviour across a range of scenarios before making it available to end users.

After Creation

Once the agent is created, you land on its details page. From here you can review the full configuration, run tests, deploy the agent, and return to make iterative edits as your requirements evolve.